Note: Descriptions are shown in the official language in which they were submitted.
DEMANDE OU BREVET VOLUMINEUX
LA PRESENTE PARTIE DE CETTE DEMANDE OU CE BREVET COMPREND
PLUS D'UN TOME.
CECI EST LE TOME 1 DE 2
CONTENANT LES PAGES 1 A 95
NOTE : Pour les tomes additionels, veuillez contacter le Bureau canadien des
brevets
JUMBO APPLICATIONS/PATENTS
THIS SECTION OF THE APPLICATION/PATENT CONTAINS MORE THAN ONE
VOLUME
THIS IS VOLUME 1 OF 2
CONTAINING PAGES 1 TO 95
NOTE: For additional volumes, please contact the Canadian Patent Office
NOM DU FICHIER / FILE NAME:
NOTE POUR LE TOME / VOLUME NOTE:
METHODS AND SYSTEMS FOR IDENTIFYING MARKERS OF COORDINATED
ACTIVITY IN SOCIAL MEDIA MOVEMENTS
[0001] Continue to paragraph [0002].
[0002] Continue to paragraph [0003].
BACKGROUND
1. Field
[0003] The present disclosure relates to methods for classifying at least one
contagious phenomenon
propagating on a network.
2. Description of the Related Art
[0004] Internet-based technologies, and the manifold genres of interaction
they afford, are re-architecting
public and private communications alike and thus altering the relationships
between all manner of social
actors, from individuals, to organizations, to mass media institutions.
Internet technologies, have enabled
shifts in methods and practices of interpersonal, communication. Many-to-many
and social scale-spanning
internet communications technologies are eliminating the channel-segregation
that previously reinforced
the independence of classes of actors at these levels of scale, enabling (or
more accurately in many cases,
forcing) them to represent themselves to one another via a common medium, and
increasingly in ways that
are universally visible, searchable and persistent.
[0005] Online readers typically navigate hyperlinked chains of relate d
stories, bouncing between numerous
websites in a hypertext network, returning periodically to favored starting
points to pick up new trails.
Hyperlinks result from a combination of choices, from those made by
individual,
1
CA 3068264 2022-08-03
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
autonomous authors to those made programmatically by designed systems, such as
permalinks,
Site navigation, embedded advertising, tracking services, and the like. Human
authors practice
the same kind of information selectivity online.that They do offline, i.e.,
what authors (including
those representing organizations) write about and link to reflects, somewhat
stable interests,
attitudes, and social/organizational. relationships. The structure of the
network. formed by these
.hyperlinks is-a product ofttiese Choices,-and thus large.scale
regularitieS_in choices will be evident
in macro-level structure. This structure will thus bear the mark of individual
preferences and
characteristics of designed systems and allows a kind of "flew map" of how the
Internet channels
attention to...online reseurces. Discriminating among types of links, and the
ability to select
categories of those which represent Author choices, allows structural
analytics to. discover
similarities among authors. Errors, randomness, or noise in linking at the
individtiatlevel has
local, 'independent causes, and does not bias large-scale macro patterns.
100061 Thus, in order to understand and leverage the online information
ecosystem, there remains
a need for systems and methods for structural analyties aimed at identifying
clusters of online
readers and influential authors, discovering how they drive traffic to
particular online resources
and leveraging that knowledge across various applications ranging from
targeted. advertising and
communication to expert identification, and the like. This need includes a-
need for understanding
the role of structures and similarities among authors and readers: in
situations involving
phenomena that follow a pattern of contagion, Leõ where an item of interest,
:niches a news story,
a political topic, a product, an item of entertainment content, or the like,
initiates with a single
point or a Small group, then spreads and grows through the network. Predicting
the pattern of
spread or contagion, the parties who will take interest in,..be involved with,
or be influenced by a
particular item, and the like may have great value in. a range of
applications; accordingly, a 'need
exists for methods and systems that assist in or enable such prediction of the
behavior of
-contagious phenomena.
SUMMARY
100071 In embodiments, methods and systems generally include determining
coordinated
_activity in social media movements on a social media. channel. The method
includes identifying.
a. plurality of markers of coordinated activity through analysis of campaign
signals from the
social media Movements. The tneth-od includes configuring a data structure of
the plurality of
markers for a social media campaign on a social media channel. The plurality
of markers
includes a network dimension for representing how accounts are connected, a.
temporal
dimension for representing patterns of messages over time, and a semantic
dimension for
representing a diversity of topics and meanings of the social media movements.
The method
also includes analyzing the campaign signals indicative of the coordinate
activity of the social
2
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
media movements in the social media campaign including determining tiSerS-
Within the social
media Campaign, determining clusters of users that makeup the social media
campaign. and
determining relationships:between the users participating in the social media-
movements, and
determining propagation patterns across clusters of users of the social media
campaign.
100081 In embodiments, identifying the plurality of markers includes
evaluating a degree to
which thecoordinated activity of the social media CaMpaigh is concentrated in
the clusters of
users. In embodiments, the coordinated activity of the social media campaign
is determined
from user actions within The Social media movements in the social media
campaign. In
embodiments, identifying the plurality of markers includes evaluating a degree
lo,which the
coordinated attlififY of the -sOcial media campaign is distributed among the
clusters 0:1 Users. In
embodiments, the plurality Of markers includes a day peakedness marker that
indicates a
percentage of the coordinated activity of the social media campaign that take
place on a day
identified as most active of the social media campaign. In embodiments, the
plurality of
markers includes_ &commitment signal that is computed by averaging a number of
subsequent
participation actions for each of a plurality of participants in the
coordinated activity of the
'social media campaign. In embodiments, the plurality of markers includes 'a
post regularity
commitment signal that represents a deviation of commitment to participation
by a user from
natural human attention patterns, In embodiments, identifying theplurality of
markers includes
determining a semantic diversity score for the coordinated activity of the
social media campaign
by assigning messages in the campaign to topics and calculating a diversity of
the topics on a
topic distance scale that facilitates determining the semantic diversity
score. In embodiments,
identifying the plurality of markers includes, computing temporal alignment of
campaign-related
actions for users in the campaign by comparing temporal sequences of campaign-
related actions.
In embodiments, identifying the plurality of markers includes computing
semantic diversity over
time to identify to-occurring topics in the social media campaign. wherein a
relatively small
value of the semantic diversity score..is configured to be indicative of
fabricated campaigns,
wherein a relatively large value of the semantic diversity score is configured
to be indicative of
sparnbots, and wherein a semantic diversity score having a value in-between is
indicative of
normal human activity,
100091 In 'etribOdimenta methods and systems generally include a computer
system for
determining coordinated activity in social media movements on a social media
channel. The
system includes a user interliee that configures a social media campaitm on
one or more social
media -channels and that communicates via a network. The system includes
acomputing device
that identifies a plurality of markers of coordinated activity through
analysis of campaign signals
from the social media movements and that configures one or more data
structures containing the
3
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
plurality of markers for the social media campaign on one or More social media
channels.- The -
plurality of markers includes a network dimension: for representing how
accounts are connected,
a temporal dimension for representing patterns of messages over time, and a
semantic dimension
for representing a diversity of topics and meanings of the social media
movements. The analysis
of the campaign signals indicative of the coordinated activity of the social
media movements in
the social Media 'eampaign includes determining users within the social Media
campaign,
determining Clusters of users that make up the social media campaign and
determining
relationships between the. users participating in the social media movements,
and determining
propagation patterns across Ousters of users Of thelawialmediaeampaign. The.
system includes
a storage system that Stores one or More of the data- StrattnreS Containing
the phirality of markers
for the social media campaign on one or more of the social media channels The -
SYStem
includes
100101 a processing system that executes computer-readable instructions that
cause the
processing system to: receive a request from an external system about the
coordinated activity of
the campaign signals from the social media movements; retrieve at least a
portion of one or
more data structures containing. the plurality of market* for the social Media
campaign On one or
more of the social media channels; and transmit contents of at least a portion
of the -analysis to
the user interface that displays at least a portiortof the plurality of
markers indicative one of
coordinated activity and normal human activity
100111 in embodiments, identifying the plurality of markers through analysis
of campaign
signals includes evaluating a degree to which the coordinated activity of the
social media
campaign is .concentrated in the clusters of users._ In embodiments, the
coordinated activity of
the social media campaign is determined from useractions within the social
media movements
in the social media campaign. The coordinated activity includes -a relatively
large number of
accounts onone=or more of thesoCial media channels controlled by a relatively
smadtannber of
coordinated entities resulting=in a. relative lack of diversity of similar
accounts on One or more
social medial channels controlled by uncoordinated users. In embodiments,
identifying the
plurality of markers through analysis of campaign signals includes evaluating
a degree to which
the coordinated activity of the social media campaign is distributed. among
the clusters of users.
1001.21 In embodiments, the plurality of markers includes 4 day peakedness
marker that indicates
a percentage of the coordinated activity of the social media campaign that
take place onady
identified as most Active of the social media campaign. In embodiments, the
plurality of
indicators includes a commitmentsignal that is computed by averaging-a, number
of subsequent =
participation actions for each of a plurality of participants in the
coordinated activity of the
social media campaign. In embodiments, the plurality of indicators includes a
post regularity
4
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
commitment signal that represents a deviation of commitment to participation
by a user from
natural human attention patterns. In embodiments, identifying the plurality of
markers through
analysis of campaign. signals indudes determining a semantic diversity score
for the coordinated
activity of the social media campaign. Determining a semantic diversity score
includes
assigning messages in the campaign to topics and calculating a diversity of
the topics on a topic
distance scale that facilitateS determining the semantic diversity score. In
embodiments,
identifying the plurality of markers through analysis of campaign signals
includes computing
temporal alignment of campaign-related actions for users in the campaign by
comparing
temporal sequencesiotetunpaign-relatad actions. In embodiments, identifying
the. plurality of
Markers through analysis of campaign signals includes computing semantic
diversity over time
to identify co-occurring topics in the social media campaign. A relatively
small value of the
semantic diversity score is Configured to be indicative of fabricated
campaigns, a relatively large
value of the semantic diversity score is configured to be indicative of
spambots, and a semantic
diversity score having avaltic in-between is indicative of normal human
activity,
1001.31 In an aspect of the. disclosure, methods and systems are provided
that: allow
characterization of structures and features of networks. Such as online
.networks of creators and
'consumers of items of content, in turn enabling prediction course of action
of actors in :such
networks and the flow of items, such asitems of content, through such
networks, including the
growth and spreading of contagious phenomena.
(0014] In an aspect of the disclosure, a computer-readable storage medium with
an executable
program stcired thereon, wherein the program instructs a processor to perform
the-steps Of attentive
clustering and analysis, may include constructing an online author network,
wherein constructing
the online author network includes selecting a set of source nodes (S), a set
of outlirik: targets (1)
from at least one-seleeted type oftyperlink, and a set of edges (E) between S
and T defined by
the at least one selected-typo of hyperlink from S to T during a -specified
time period; deriving a
set of nodes, r, .by any one of or combination of a.) normalizing nodes in T.
optionally to a
selected level of abstraction, b.) using lists of target nodes for exclusion
("blacklists"), and e.)
using lists of target nodes for inclusion ("whiteliste); transforming the
online author network into
a. matrix of source. nodes in S linked to targets in T; partitioning the
online author network into at
least one set 'of source nrideS -Witit:a Sittilar linking history to form an
attentive cluster and/or at
least one :set of outlink targets with a similar Citation profile, to form ati
Otitlink bundle; and
optionally, generating a graphical representation of attentive clusters and/or
oudink bundles in the
network to enable interpretation of network features and behavior_ and
calculation ofcomparati ye
statistical measures across the attentive clusters and outl ink bundles;
wherein at least one clement
of the graphical representation depicts a measure of an extent of a type of
activity within the
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
network; and measuring frequencies of links between attentive clusters and
outlink bundles
enabling identification and measurement of large-scale regularities in the
distribution of attention
by online authors across sources of information. The element of the graphical
representation may
use at least. one pf size, thickness, color and pattern to depict a type of
activity. Attentive clusters
and their constituent nodes may he differentiated in the graphical
representation by at least one of
color (including hue. 'intensity and saturation),. a =shapelincluding 2D or 3D
representations), . a
=geometric arrangement, a 'shading, a transparency and a size. The size of the
object representing
the clustered nodes in the graphical. representation:may:correlate with a
metric. The nodes, targets,
and edges may be collected from public and private sources of information.
Constructing the
Matrix. may include applying at least one threshold parameter from the group
.consisting of:
maxnodes, targetmax, nodemin, targetmin, maxlinks, and linktnin. Constructing
the matrix may
include applying a minimum threshold tbr the number of included mxies that
must link to a target
to:qualify it for inclusion in the matrix. Constructing the matrix may include
applying a minimum
threshold. for the number of included targets that must link to a node to
qualify it for inclusion in
the matrix. The matrix may be a graph .matrix. The method may further include
applying any
lists specifying inclusion or exclusion of particular nodes.
100151 it should be understood that, except where context prevents, the term
"author," as used
herein, should be understood. to encompass human and non-human creators and
editors. of content
(including, without .limitation, text, images, video, tweets,. animations,.
multimedia and any
combinations. or other types of content and. including, without limitation,
original content,
derivative Works, commentary, analysis, and other genres of content) that can
be consumed (e.g.,
read or viewed) by others, such as readers or viewers in a network,.
1001.61 In an aspect of the disclosure, a method of using attentive clustering
to steer a further data
collection process may include partitioning an online, author network into at
leastone set of source
nodet:::With a similar linking history to forth an attentive duster and at
least one set of outlink
targets with a similar citation profile- to form an outlink bundle, and
collecting .clickstream data
for the source nodes of the attentive cluster.
100171 man aspect of the disclosure, a method of using attentive clustering to
steer a -further data
collection process. may include partitioning an online author network into at
least one set of source
nodes with a similar linking history to fOrni.art attentive cluster and at
least- one set Of outlink
targets with a similar citation profile 1.05Ortkan ontlinkbtindle, and
collecting clickstream data
for the target nodes of the outfit* bundle.:
1041181 In an aspect of the disclosure, a method of using attentive clustering
to steer-a further data
collection process may include partitioning an online author network into at
least one set of source
nodes with a similar linking history to form an attentive cluster and at least
one set of outlink
6
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
targets with a similar citation. profile to form an outlink bundle, and
collecting survey data for the
Source nodes of the attentive cluster.
100191 in an aspect of the disclosure, a method of using attentive clustering
to steer a further data
collection process may include partitioning an online author network into at
least one set of source
nodes with a similar linking history to thrift an attentive cluster and at
least one set of outlink
targets with a similar citation profile to form an outlink bundle, and
collecting survey data for the
target nodes of the outlink bundle.
100201 In an aspect of the disclosure, a method of using attentive. clustering
to steer a -further data
collection process may include partitioning an online author network into at
least one set of source
nodes with a similar linking history ari form an attentive cluster: and at
least one set of outlink
targets with a- similar citation profile to form an outlink bundle, and
collecting geo-location data
for the source nodes of the attentive cluster.
100211 In an aspect of the disclosure, a method of using attentive clustering
to steer a further data
collection process may include partitioning an online authornetworkinto at
least one set of source
nodes with a similar linking history to form an attentive cluster and at least
one set of outlink
targets with a similar citation profile to form an outlink bundle, and
collecting geo-location data
for the target. nodes of the outlink bundle.
100221 man aspect of the disclosure, a method of metadata tag analysis to
facilitate interpretation
of an attentive cluster may. include partitioning an online author network
into at least one set of
source nodes with a similar linking history to form an attentive cluster and
at least one set of
outl ink targets with a similar citation. Profile to form an outlink bundle,
collecting a metadata tag
associated with the source nodes in the attentive cluster, and performing a
differential frequency
analysis on the metadata tags- that are: associated with the attentive
cluster. The method may
further include sorting cluster focus scores on a plurality of the .metudata
tags.
100231 In an aspect of the disclosure, a method of metadata tag analysis to
facilitate interpretation
of' an attentive cluster May include partitioning an online author network
into, at: least one set of
source nodes with a similar linking history to form an attentive cluster and
at least one set of
outlink targets with a similar citation profile to form an outlink bundle,
collecting a metadata tag
associated with the source nodes in the attentive cluster, and performing a
differential frequency
analysis on the metadata tags that are associated with the. outlink bundle.
The method May further
include sorting cluster focus scores on a plurality of the metadata tags.
100241 In an aspect of the disclosure, a method may include partitioning an
online author network
into at least one set of source nodes with a similar linking; history to .form
an attentive cluster and
at least one set of ()tank targets with a similar citation profile to form an
outfit*. bundle, forming
a density matrix of the attentive cluster and the outlink bundle, determining
where there is a. higher
7
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
density in the density matrix than chance would prediceand identifying
patterns of influence of a
block of web sites on a block of authors by analyzing the higher density area
of the density matrix,.
100251 In an. aspect of the disclosure, .a method of macro measurement of link
density may include
constructing an online author network, wherein, constructing the online Author
network comprises
-selecting a set of source nodes (1S),. a set of outlink targets (T), and a
set of edges (I) between S
and T defined by the at least 'One selected type of hyperlink from. S to T
during a specified time
period, deriving a set of nodes, r, by normalizing nodes in T, transforming-
the online author
network into a matrix of source nodes iOS linked to targets in and
collapsing the matrix to
aggregate link measures among clusters Pfspumes and clusters of targets.
Thei:.aggregated link
Measure may be at least one Of a count of number of nodes in source cluster S
linking keit*
member of target set T. a density calculated by dividing counts by the.
product of the number of
members in S and thenumber of members in T; and a standard score that is a
standardized measure
of the deviation from random chance for counts across each source node-outlink
target crossing
in the density matrix.
100261 In an. aspect of the disclosure, a method. may include partitioning an
online author network.
into at least one set of source nodes with .a similar linking history to form
an attentive cluster and
at least one -set- of outfit* targets with a. similar citation profile to form
.an outlink bundle, and
associating the attentive cluster with a real world group. of people.
100271 in an aspect of the disclosure, a method of multi-layer attentive
clustering may include
partitioning a multi-layered social segmentation into at least one set of
source nodes with a similar
linking history to font an attentive cluster and at least one. set of outlink
targets with a similar
citation profile to form an outlink bundle, and monitoring-at least one of the
attentive cluster and
the outlink bundle on at least one layer of the social segmentation. The
social segmentation may
be an online social media author network. Monitoring may be tracking the
growth of= attentive
.cluster overtime, The method May further inolude examining a source node
associated, with a
specific player in the attentive cluster in order to deteenine a
characteristic. The monitoring may
be used to identify a group of people who are susceptible to a message and
track -downstream
activities in response to the message.
100.281 In an aspect of the disclostire,a Method May ieclude partitioning an
online author network
into at least one set of source nodes with a similar linking history to form
an Atteefteeeluster and
at least one set of outlink targets with a similar citation profile to form
ari outlink bundle, and
analyzing the attentive cluster Over time to depict changes in a linking
pattern of the attentive
-cluster over a time period. Theoutlink bundle may be &list of semantic
markers. The semantic
marker may be at least one of a text element, a post, a twee, an online
content, and a inetadata
8
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
tag. Analyzing may itivOlve tracking a semantic marker or set of semantic
markers across one or
more attentive clusters within the online author network.
100291 In an aspect of the disclosure, a method.may include partitioning an
online author network
into at least one set of source nodes with 4 similar linking history to form
an attentive cluster and
at least one set of outlink targets with a similar citation profile to form an
outlink bundle, and
calculating a set of cluster focus index (071), scores for the attentive
.cluster, wherein the C171
represents the degree to which a particular outlink target is
disproportionately cited by members
of a particular attentive cluster as compared to the 'average citation
frequency for all nodes in S.
At least one source node:may .he a. high attention source node. The method may
further include
automatically placing-an 'Advertisement at the particular outlink target,
100301 In an aspect Of the disclosure, a method may include partitioning an
online author network
into at least one set of source nodes with a similar linking history to tbrm
an attentive cluster and
at least one set of outlink targets with a similar citation profile to form.
an outlink bundle, and
generating a graphical representation of attentive clusters and/or out! ink .
bundles in theitetwork
to enable interpretation of network features and behavior and calculation of
comparative statistical
measures across the attentive clusters and Win* bundles,, wherein at least
One element of the
graphical representation depicts a measure of an extent of a type of activity
within the network.
The method may further include further segmenting the network using at least
one of a text, an
item of online content, a link; and an object. The source node in the
graphical representation may
he represented by an individual dot. The size of the dot may be determined
based on the number
of other source nodes that link to it.
100311 In an aspect of the disclosure, a -method may include:partitioning an
online author network
Into at least one set of source nodes with a similar linking history to form
an attentive cluster and
at least one set of outlink targets with L.:similar citation profile to form
an oudink bundle,
calculating a set of cluster focus index (CF1),:40Ofes (a% tot' the attentive
cluster, wherein the
CFI represents the degree. to which a particular (Milli* target.is
disproportionately cited by at least
one source node of a. particular attentive cluster, and generating a graphical
representation of
-attentive clusters and/or ()Wink boodles in the network, wherein at least one
element of the
graphical representation depicts a measure of an extent of 4 type of activity
Within the network,
wherein the higher the C.Ftfcbre, the higher the outlink target appears along
at least one axis of
the graphical repmsentation.
100321 In an aspect of the disclosure, a method of attentive clustering may
include defining a
semantic bundle, searching a plurality of candidate nodes tbr items in, the
bundle. in order to
generate a relevance metric for use in selecting high-relevance online
authors, partitioning the
online author network into at least one set of source nodes with a similar
linking history to form
9
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
an attentive cluster and at least one. set of outlink targets with a similar
citation profile to form an
outlink bundle, and calculating metrics with across clusters for items in the
semantic bundle.
100331 In an aspect of the disclosure, a method.may include partitioning an
online author network
into at least one set of source nodes with a similar linking history to form
an attentive -Ouster and
at least one set of outlink targets with a similar citation profile to form an
outlink bundle, and
generating a graphical representation of link targets, semantic -events, and
nede-asSoeiated
metadata scattered in an x-y coordinate space, wherein the dimensions of the
graph are custom-
defined using sets of attentive clusters grouped to represent substantive
dimensions of interest for
a particular .analysis.
100341 In an aspect, a computerized search method may include presenting, to a
user, a computer
interface for specifying one or more search terms for a search query,
presenting. at least one
selectable item corresponding to at least one Of an M score and a CFI score
filter for the search
.query, generating an amended search query based on a selected item, and
performing a search
using the amended search query. The search may be of the Internet, The search
may .be of #
document-corpus. The search may be of a CO!filtered set of clusters within an
online network.
The search may be of a set of nodes having an M score greater than a
threshold.
.100351 CFI may represent the degree to which an event, characteristic or
behavior
disproportionately occurs in a particular .cluster, or a particular cluster-,
relative to a network,
preferentially manifests an event, characteristic or behavior. M score may be
calculated using the
formula IV score.vo.unt (alpha)+CFI ( I -alpha) [normalized I to In where
count is the overall.
number of members on a cluster focus map that have engaged with a target.
10036l in an aspect, a computerized search method may include presenting,.to a
user, a computer
interface for specifying one or more search terms for a search query,
presenting, to the user, a
computer interface for selecting content to search with the search terms,
wherein the content is
taken from an online creator network partitioned into at least-one set of
source nodes with a similar
linking history to form an attentive cluster and. at least one set .of outlink
targets with _a similar
citation profile to form an outlink. bundle, and performing a search of the
selected content using
the.search query.
[0071 in an aspect, .a method to iteratively reduce the scale of a network to
its most influential
eore ermunanities and obtain a -sub-graph of maximally connected sub-actors
may include
assigning ,:a Variable, Keorr, to each individual member of the network.,
where Kan relates to a
minimum connectedness based on the number of other nodes in the network to
which the
individual, is connected, removing inactive individuals and individuals with
few followers from
the network, temporarily removing certain individuals with a large number of
followers for later
:re-joining, restricting the remaining individuals iteratively by removing
individuals with the
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
lowest = Kcort vette& first, then removing individuals with the next highest
Ktorr values until a
threshold is reached, wherein the threshold is at least one of a number of
individuals removed, a
number of individuals remaining, and a .Kart value, and re-joining the.
temporarily removed
individuals.
100381 in an aspect, a self-service tool to construct a social media map may
include an automated
process (e.g., hot) that harvests data (e.g., nodes) and maps the data to one
or more
clusters/segments, a processor that provides elasterisegment labels and. CFI
scores for the
clusters/segments, and an interface that enables user broWSing of
clusters/segments and the map,
tagging nodes, and re-groupingft-e-labeling_of dusters/segments. The automated
process may also
be capable of: -automatically refreshing the social Media map based on using a
relevance score -.for
nodes in the map, positively_ or negatively weighting. at least one cluster
based on a CFI score
calculation to include positively weighted nodes and exclude negatively
weighted nOdes from the
map, filtering out unwanted nodes, obligatorily including nodes that were not
clustered in a first
version of-the social media. map, crowd-sourced information regarding nodes
and/or links that
drives nodes to bundles, processing social media map usage data for
trends/indicators, wherein
the Usage data relates to one or more of what is ignored, what is further
explored, what is used,
how are clusters arouped, what namellabel is assigned 'tea cluster, what color
is used fora cluster,
what order/position is.the cluster placed in a report and wherein, nodes
preferentially interacted
with are weighted more heavily, and user-contributed data as metadata for the
social media map.
100391 In an aspect, a method of strategic messaging may include generating .a
list of targets in a
networklehtster/Segment, filtering the list by a criteria to limit Whom to
message -in the
network/cluster/Segment in order to maximize the impact of the message on the
cluster/segment,
wherein the filter is at least one of CFI score, M score, number of followers,
following status,
follower status, number of mentions/re-tweets, number of distinct mentions,
status of exposure to
.content, status Of exposure to content that has already peaked, footprint,
and number Of
tweets/publication frequency, and ranking the list by the filtered criteria.
100401 In an aspect, a method of strategic network building may include
generating a list of targets
in a networldelusterisegment,=wherein the list is generated. using at least
o.ne:.of-CF.1õ M score, # of
-followers., mentions/Iv-tweets, distinct 111e01iotts, and number -of tweets,
and following the targets.
OM hi an aspect,--a Method Of calculating M score may include calculating a
cluster focus index
score based on a degree to which a target disproportionately occurs in a
particular cluster, or a
particular cluster, relative to a netWorle preferentially engages with a
target, determining an
overall, number of members of the cluster, or -network that have engaged with.
that target, and
calculating an M score based on the formula: count plus CFI, wherein count is
the overall number
of members of the cluster that have engaged. with that target.
[0042] In an aspect, an M score filter for a list of targets may include
taking a cluster focus index (CFI)
score based on a degree to which a target disproportionately occurs in a
particular cluster, or a particular
clyster, relative to a network, preferentially engages with a target, and
providing a slider to indicate an M
score, wherein the M score is based on the formula: count (alpha)+CFI (1 -
alpha), wherein count is the
overall number of members of the cluster or network that have engaged with
that target, and wherein the
slider is used to indicate the value of alpha between 0 and I.
[0043] In an aspect, a method of strategic ad placement may include generating
a list of targets in a
network/cluster/segment representing linkages in a social media environment,
filtering the list by a criteria
to limit the targets in order to maximize the impact of the ad on the
network/cluster/segment, wherein the
filter is at least one of CFI score and M score, ranking the list by the
filtered criteria, and providing an
interface to launch an ad campaign to place ads directly from the environment
representing the linkages to
the target/website. Ad placement may be done via integration with various
products, such as TwitterTm
sponsored tweets, FacebookTM ad exchange, GoogleTM Adsense/Adwords, and third
party online ad
networks. The method may further include tracking interaction with the ad
across social networks.
[0044] In an aspect, a method for using cosine similarity to determine the
relationship between one or more
clusters may include for each cluster, building a vector based on the CFI
scores calculated for a number of
items, plotting the vectors in a 3D vector space, determine the cosine of the
angle between the vectors as
an indication of the relationship between, the clusters, and when a
relationship is identified between clusters
based on the cosine, automatically labeling the clusters with the same label.
If the cosine is small, the
confidence that there is a high degree of similarity is high.
[0045] In an aspect, a method may include publishing a map of content as a
widget, and tracking interaction
with the content in the widget to obtain behavioral data about a user of the
map.
10046] In an aspect, a method may include publishing a map of content as a
widget, tracking interactions
with the content in the widget to obtain behavioral data about a user of the
published map; and analysing
the behavioral data in order to at least one of suggest content, track network
evolution, modify the network
in strategically valuable ways, and measure the success of an ad campaign.
[0047] These and other systems, methods, objects, features, and advantages of
the present disclosure will
be apparent to those skilled in the art from the following detailed
description of the preferred embodiment
and the drawings.
[0048] References to items in the singular should be understood, to include
items in the plural and vice
12
CA 3068264 2022-08-03
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
versa, unless explicitly stated otherwise or clear from the text. Grammatical
conjunctions are
intended to express any and all disjunctive and conjunctive combinations of
conjoined elauses,
sentences, words, and the like, unless. otherwise stated or clear from the
context
BRIEF DESCRIPTION OF Ttit FIGURES
100491 The structures, methods, systems, inventions and the ibllowing detailed
description of
certain embodiments thereof may be understood by reference to the following
figures:
100501 FIG. 1 depicts a process flow for attentive clustering.
100511 FIG. 2 depicts a social network map in the form Of &proximity duster
map.
tooszi F1(1.- 3 depicts a social network map in the form of a proximity
cluster map highlighting
attentive clusters of liberal and conservative U.S. hloggers, and BritiSh
bloggers.
100531 FIG. 4. depicts a social network map in the form of &proximity duster
map tbcused on
environmentalists, feminists, political bloggers, and parents.
100541 FIG. 5 depicts a social network map in the form of a proximity cluster
map with a cluster
relationship identified.
100551 FIG. 6 depicts a social network map inAlieform. of a proximity duster
map with a bridge
blog identified.
100561 FIG. 7 depicts a flow diagram for attentive clustering.
100571 FIG. 8 depicts a Political Video Barometer valence graph.
100581 FIG. 9 depicts a graph of CFI scores.
100591 FIG. 10 depicts a graph of CFI scores.
100601 AO, LI depicts a hi-polar valence graph of link targets in the Russian
blogosphere.
100611 :Ka 12 depicts an interactive burstmap interface.
100621 FIG..13 depicts a valence graph of outlink targets organized by
proportion of links from
ifiberldvs:..conservative bloggers.
:-E09031 FIG. 14 depicts a flow diagram relating to social media maps.
100641 FIG. 15 depicts a flow diagram relating to refreshing social media
maps.
100651 FIG: 16 depicts a flow diagram relating to social media maps.
100661 FIG, 17 depicts formation ofa ranked target list.
00671 FIG. 18 depicts Peakedness vs. Commitment by Time Range for two sets of
hashtags.
100681 FIG. 19a Ogyjets Peakedness vs. Commitment by Subsequent Uses.
100691 FIG. 1917-depicts Peakedness vs. COMMitment by Commitment by Time
Range.
100701 146: .20 depicts a .distribution of mention-weighted normalized
concentration by topic.
100711 FIG..21 depicts.a distribution of Cohesion bytopie.
100721 FIG.. 22a depicts a chronotope of the timetro29 hashtag.
100731 FIG. 22b depicts a chronotope of the fisamara hashtag.
13
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
100741 .FI(3. 22c depicts a chronotope of the #iRu hashtag.
100751 FIG.-23-depiets a social media map platform user flow.
100761 FIG. 24 idepiets.a recent activity page for asocial media map platform,
100771 FIG. 25 depicts a recent activity page fora social media map platform.
100781 FIG. 26 depicts an overview page for a social media map platform
100791 FIG. 27 depicts an Inter etive map for a social media map platform,
100801 FICL.28 depicts an overview page for a social media map platform.
100811 FIG. 29 depicts an influencers page for a social media. map platform.
10041 'FiCi,.:30:depicts an influencer detail fOra,social media map platform.
100831 depicts.a.conversation leaders page for a social media map
platform.
100841 FIG. '32 depicts. a tweets page for a social media map platfoms.
100851 FIG.33 depicts a websites.page fbr a social media map platform.
100861 FIG..I.34 depicts a key content page for a social media map platform.
100871 F.1(1.15 depicts a media pagc.for a social media map platform.
100881 FKI..36 depicts a terms page for a social media map plattbrin.
100891 :FIG. ..37 depicts a lists page: for a soda! media Map platform.
DETAILED DESCRIPTION
190901 The present disclosure relates to a computer-implemented method
forattentive clustering
and analysis. Attentive clusters are groups of authors who share similar
linking profiles or
collections of nodes Whose use of sources indicates common attentive behavior.
Attentive
clustering and related analyties may include- measuring and visualizing the
prominence and
specificity of textual elements, semantic activity; sources of information,
and hyperlink.ed objects
across emergenteategorift of online authors within targeted subgraphs of the
global Internet. The
disclosure may include a set of specialized parsers that identify and extract
online conversations.
The disclosure may include algorithms that Cluster data and mapthem into
intuitive visualizations
(publishing nodes, Wogs, tweets,. etc.) to determine emergent clusterings that
are highly navigable.
The disclosure may include a front end/dashboard for interaction with. the
clustering data. The
disclosure may include a database for tracking clustering data.. The
disclosure may include tools
and data to visualize, interpret and act upon measurable relationships in
online media. The
approach may be to segment an online landscape based on behavior of authors
over -time, thus
creating an emergent segmentation of authors based on teal behavior that
drives metrics, rather
than driving metrics based on pre-conceived lists. Because the analysis is. a
structural one, rather
than language-based, the analysis is language agnostic In an embodiment, the
segmentation may
be global, such. as of the English language blowsphere. In an embodiment, the
segmentation may
-involve a relevance metric for every node based on semantic markers and a
custom mapping of
14
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
high-relevance nodes. The disclosure enables identifying influencers, such as
who is authoritative
about what to whom.
100911 One method of obtaining attentive clusters may involve construction of
a bipartite matrix,
however, any number and variety of fiat or hierarchical clustering algorithms
may be used to
obtain an attentive cluster in the disclosure. in an embodiment, a set of
content-publishing source
nodes ("authors") may be selected based on a chosen combination of linguistic,
behavioral,
Semantic,. network-based or other criteria. A mixed-mode network may be
constructed,
comprising the set .S. Of all source nodes, the set T of all outlink targets
from selected types of
hyperlinks, and the:.s.e(0:otedges between them defined by the selected type
or types of links
from S to T found dmittga specified time period. A matrix, such as a bipartite
graph. matrix, may
be constructed of. source nodes inS linked to targets in V. derived by any
combination of a.)
normalizing nodes in Tõoptionally to a selected level of abstraction, b.)
using lists of target nodes
for exclusion ("blacklists"), and c.) using lists of target nodes for
inclusion ("whitelists"). The
.matrix may represent a two-mode networkfor actor-event netweNthin associates
two completely
different categories of noes, actors and events, to build a network. of actors
through their
participation in 'events or affiliations. In embodiments, the matrix is, in
effect an affiliation matrix
Mall 'authors with. the things that: they link to, wherein the patterns of
their linking may be used to
do statistical clustering of their nodes.
100921 The matrix may be processed according to user-selected parameters, and
clustered in order
to perform one or more of the following: 1.) partition the network into sets
of source nodes with
similar 'linking histories ("attentive clusters"); 2.) identify sets. of
targets (linked-to websites or
objects) with similar citation profiles. ("outlink bundles"); 3.) calculate
comparative statistical
measures across these partitions/attentive clusters; 44. construct
visualizations to aid in
interpretation of network features and behavior; 5.) measure frequencies .of
links between attentive
clusters and otttlink bundles, allowing identificatien and measurement of
large-scale regularities
in the distribution of attention by authors across sources of intbrmation, and
the like. An arbitrary
number and variety of flat or hierarchical clustering algorithms may be used
to partition the matrix,
and the results may be stored. in order to select any solution for output
generation. The resulting
outputs (measures and visualizations) May provide .novel, unique, and useful
insights for
determining influential, authors and Websites, planning comm.unications
strategies, targeting
online advertising, and the like.
100931 in an embodiment, systems and methods for attentive clustering and
analysis may be
embodied in a computer system comprising hardware andariffware elements,
including local or
network access to a corpus of chronologically-published interrtet.dataõ such
as blog posts, RSS
feeds, online articles, lwitterTM "tweets," Facebookm postings, italthe like.
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
190941 Referring to FRi. 1, attentive-clustering and analysis may include: 1.)
network selection
1.02. 2.) partitioning 104, which may include two-mode network clustering in
this embodiment,
and .3.4 visualization and metrics output 10$. Network selection 102 may -
include at least twn
operations: a.) node selection 110, and b.) link- selection 112. Optionally, a
third. may be applied
in which network analytic operations are used to further specify the set of
source nodes under
consideration for clustering. For example, the operation .may be filtering.
Filtering may be
technology-based, blacklist-based, whitelist-based, and the like.
100951 In an embodiment, nodes. may be URI,s, at Which chronologically
published streams or
elements of content may be available. An initial set: containing any number of
nodes may be
selected bated on any combination Of node-level characteristics and/or
calculated = relevance
scores. Regarding node-level characteristics, there may be anumber of
different kinds of nodes
publishing content online, such as weblogs (blogs), online media sites (like
newspaper webSites),
microblogs (like Twitterm), forums/bulletin boards (like http://www.biology-
ortline.orgibiology-
forum feeds (like and the like. In addition to different technical
genres of node,
nodes may differ according to an arbitrary number of other intrinsic or
extrinsic node-level
characteristics, such as the hosting platform (e.g., BIOOpot, Livejoumal), the
type of content
published (text, images, audio), languages of textual .'content (e.g., Frenth,
Spanish), type of
authoring. entity (individual, group, corporation, NGO.,.government,_.online
content aggregator,
etc.), fiNueriey or regularity of publication (:14y, regular, monthly,
bursty), network
characteristics (e.g., central, authoritative, A-list; isolated, un-linked,
long-tail), readership/traffic
levels, geographical_ or political location of authoring entity or feats of
its concern (e.gõ Russian
language, Russian Federation,. Bay Area Calif.), membership in a particular
online ad distribution
network. (e.g., BLOGADS, GOOGLErm ADSENSE),-third-party categorizations, and
the like.
190961 To support node selection 110 based on relevance to particular issues
or actors, or
relevance-based Made selection 11.0,fist,sof relevance markers may he used to-
calculate composite
scores across nodes:. These lists may include such items as key words and
phrases, semantic
entities, full or partial URI,s, meta tags embedded in site code and/or
published documents,
_associated tags in third-party collections (e.g.,- DELICIOUS tags), and thc
like. For example, tags
-may be collected automatically, such as by "spidering" sites for meta
keywords. The corpus of
internet data may be seanned and matches on list elements tabulated for each
node. A number of
methods may be used to calculate a relevance score 'based on. these match
counts. In an
-embeditnent, relevance scores may be calculated by calculating individual
index scores- for text
matches (T), link matches (1), and metadata matches (N), and then summing-
them. These
individual index scores (1) may be calculated for each node by 'scanning all
content published by
a. node during a specified period of 'time using a list of j relevance
markers:
16
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
I=Suri*Xi*W1ntffievezYt2 . (xj*wj)/ti), where x is the number of matches for
the item, w is a
user-assigned weight (a scale Of '1 to 5 is typical), and t is the total
number of Rein matches in the
scanned corpus. In an example, an initial set of source nodes may include the
100,000 Russian
language weblogs most highly cited during a particular time frame. In another
example, the initial
set may include the 10,000 English language weblogs with the highest relevance
scores based on
relevance marker lists associated with thepolitical issue of healthcare, In
another example, the
initial set may include all nodes,by Indian and Pakistani authors in whatever
language that have
published at least three times within the past Six months.
100971 With respect to the link selection 112 component of network selection -
1.02, objects may
be particular unitS ehronologiLsally published content. found at a node,
such as blog 'posts,
"tweets," and the like. Links, also referred to as outlinks herein, may be
hyperlink URts found
within a node's source HTML code or its published objects. Many kinds of links
exist, and the
ability to choose which kinds are used for clustering may be a key feature of
the method. There
are links for navigation, links to arehives, links to .servers for embedded.
advertising, links in
comments, links to link-tracking services, and the like. link selection 112
may be applied to links
that represent deliberate choices made by authors, of which there may also be
many kinds. These
links may be to 'nodes (e.g., a weblog address found in a "blogroll"), objects
(e.g., particular
YOUTUBET" video embedded in a blog post), and other classes of entity, such as
"friends" and
"followers." Some node hosting platforms define a typology of links to reflect
explicitly defined
relationships, such as "friend," "friend-of," "community member," and
"community follower" in
LIVESOURNAL, or ',follower" and "following" in Twitterml, FacebookTM and the
like. In other
cases, informal :conventions, such as "blogrolls," definaiype-of link: Some of
these link types
are relatively static, meaning they are typically
availibielsvartortheinterface used by a visitor
to a node website, while others are dynamic, embeiged within-publithcd content
objects. Link
types may be parsed or estimated and stored with the link data iThete...firikt
represent different
types of relationships between authors and linked, entities, and therefore,
according to the user's
objectives, certain classes of links may be selected for inclusion. Different
sorts of links also have
time values associated with them, such as the date/time of initial publication
of an object in which
a dynamic link is embedded, or the first-detected and. most recently seen
date/time of a static link.
Links may be further selected fa clustering based on these time values.
100981 From the parameters defined for node selection 110 and link selection
112, a mixed-mode
network X 136 may be constructed, consisting of the set S of all source nodes,
the set T of all
outlink targets from selected types of hyperlinks, and the set E of edges
between them defined by
the selected type or types of links from S to Tibund during a specified time
period. The network
130 may be considered "mixed mode" because While it may be formally bipartite,
a number of
17
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
nodeairt S may also exist in T, which May be considered a violation of the
normal concept of two
mode networks. Rather than excluding nodes that may be considered either S or
T nodes, the
systems and methods of the present disclosure consider them. logically
separate, .A particular node
may .be considered a source at' attention (S) in one mode, and an object of
attention m in the
other. Before clustering, the SO of nodes may he further constrained by
parameters applied to X,
or to a one-mode subnetwork .)C consisting of the network. 130 defined by
nodes in S airing with
all nodes in T that are also in S (or at a level of abstraction under an
element in S, collapsed to the
parent node). Standard network analytic techniques may be applied to X' in
order to reduce the
source nodes under consideration for clustering. For instance, requirements
for k-connectedness
May be applied in order to limit Consideration to well-connected ,nodes
100991 In an embodiment, partitioning 104 may include: 1.) specification of
node level for
building the two-mode network, 2.) assembly of bipartite network matrix 132
using iterative
processing of matrix to conform with chosen threshold parameters, and 3.)
statistical clustering
(multiple methods possible) of nodes on. each mode, that is, source node
clustering 114 and autlink
clustering 118. Outlink clustering 118 to form an ()Wink bundle may involve
iden.tifying sets of
web sites that are accessed by the sante kinds of people.
101001 With respect to specification anode level, distinction maybe made
between "nodes"and
"objects," considering the node as a stable URL at which a number of objects
are published. This
may result in generation of a straightforward two-level hierarchy (object-
node); however, nodes
sometimes have a hierarchical relationship among each other (object-node-
metanode). Consider
the following three URLs:
101011 41.1ittrilwww.b1oghost.comi;
101021 2.) httrthvww.bloghost.comfusersijohndoetblogi; and
101.031 3.)Ittp://www.bloghoSt.comlusersijohndoelblott/0916/21/myblagpasatml-.
.101041 RCM a three-level hierarchy with a inetanode [11, node [2J, object
exists In some
eitibodinients, the node URI, May 'correspond very simply to a "hostname" (the
part of a URI,
after "http://" and before the next "P') or a hostriame plus a uniform path
element (like "ibloe;"
after the hostname). In other embodiments though, multiple nodes may exist at
pathnames under
the same hostname. Depending on the objective 011ie user, a "node level" may
be selected for
building the two-mode network, such. that seCond-inOde nodes include (from
most general to most
specific level) a.) metanodes (collapsing sub-nodes into tine) and independent
nodes, b..) child, or
sub-nodes (treated individually) and independent nodes,or e.) Objects of which
a great tnany may
exist for any given parent node). irumbodimentojtmay be possible to mix node
levels according
to a rule set based on defining levels for particular sets of nodes and.
metanodes, or on link
threSholds for qualifying objects independently. Furthermore, a node with a
webpatte URL may
18
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
-often have one or more associated "feed" LIMA, at which published content may
be available.
These =feeds are generally considered as the. same logical node as the =
parent site, but may he
considered as independent nodes. If a target URI. is not a publishing node,
but another kind of
websiteõ the level may likewise be chosen, though more levels of hierarchy may
be possible, and
typically the practical choice may be between hostname level or full patbname
j01051 .With respeet to the assembly of the bipartite network matrix 132 using
iterative processing
of the matrix 132 to conform with chosen threshold parameters:, links may be
reviewed and
collapsed (if necessary) to the proper node level as described hereinabove,
and the two-Mode
network may be built between all link sources (the initial nude set and all
target (second-mode)
nodes at the specified tvade level or ley*. Opfionally, blacklists and
whiteliSts May beOtted to.
respectively, exclude, or force inettitiionaf:Spedifie: source. or target
nodes. From this full network
data, an Nal< bipartite matrix M. in -Whitt! N is the set of final source
nodes and K is the set of
final target nodes, may be constructed according to user-specified, optional
parameters, such as
maxnodes, nodemin, maxlinks, linktnin, and the like. .An iterative
sortingalgoritam may prioritize
highly connected sources and widely cited 'targets, and then use these values
to determine which
nodes and targets from the full network data may be included in the matrix.
lailaxsources and
maxtatgets may set the maximum values for the number of elements in N And K.
Nodemin may
specify the minimum number of included targets (degree) that a source is
required to link to in
order to qualify for inclusion in the matrix. Linkmin similarly may specify
the minimarn, number
of includtal sources (degree) that must link to a target to qualify it for
inclusion in the matrix. Two
other Optional parameters, nodemax and linkmax max be used to specify upper
thresholds for
source and target degree as well. Each value (Va) in M. the. number of
individual links.from
source Ito target j.
101.061 With respect to statistical clustering in each-mode, that is node
clustering 114 and outlink
:clustering 118, there may be amitotic!. of clustering-algorithms Which may be
used. to partition the
network, including hierarchical agglomerative, divisive, k-means, spectral,
and the like. They
may each have merits for certain objectives. In an embodiment, one approach
for producing
interpretable results based on internet data. may be as follows: I .) make M
binary, reducing all
Values >0. to 1; 2.) calculate distance Matrices for M and its transpose,
yielding an NXN matrix of
distances between -sources, and a KxK Matrix of distances between targets.
Various distance
measures May be possible, but gOOd results May be obtained by converting
Pearson correlations
to distances by subtracting from 1; 3.) using Ward's method for hierarchical
agglomerative
-clustering, a cluster hierarchy (tree) may be- computed and stored for each
distance matrix.. Results
of an arbitrary number of clustering operations may-be saved in their
entirety, so that any particular
flat cluster solutions may be chosen as the basis for generating outputs.
19
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
191071 in an embodiment, the clustering algorithm may be language agnostic,
that is, forming
attentive clusters around similar targets of attention Without a constraint on
the language of the
targets. In an embodiment, clustering may make .use of metadata that may
enable the system to
know about the content of various websites without having to understand a
language. In another
embodiment, the algorithm may have a translator or work .in conjunction with a
translation
appliCatiOn in Order to find terms across publications 'of any language.
101081 Now that the. first two stages of attentive clustering, network
selection and. two-mode
network. clustering, have been described we turn to a description of
visualization and metrics
output. Any particular set_ of Cluster .sOlutions.for source nodes (an
assignment of each node to a
cluster) may be selected by the user in order to. generate one or more of the
following elasses-of
-output: 1.) per-cluster network metrics for source nodes_ 120; 2,) across
clusters comparative
frequency measures. of link, text, semantic and Other node and link-level
events, content and
features; 3.) visualizations 124 of the partitioned network combined with
these measures and other
data on node and link-level events, content and featuies;.. and 4.) aggregate
cluster metrics
reflecting ties among clusters taken as groups. Further, any particular set of
cluster solutions for
target nodes may be selected and used in combination with the set of cluster
solutions tbr source
nodes in order to generate: 1.) measures of link frequencies and densities 128
between source
clusters and target clusters; 2.) visualization 124 of the previous as a
network of nodes representing
clusters of sources and targets with ties corresponding to link densities"
28;.ari4.3.) visualizations
124 of one-mode calculated (network of target nodes) networks with:partition.
data.
101091 In one class of output* and with respect. to per-cluster network
metrics for source nodes
120, inaddition to standard network metrics for source nodes that are
generated over the entire
network, and which reflect various properties important -for determining
influence and role in
information flow, user-selected cluster solutions may be used to- generate a
set of measures for
each node, pep-cluster. These measures may represent the nodes direct and
indirect influence on,
or visibility to, each cluster, as well as its attentiveness to each duster.
For every node i, these
measures may include the following: same-in: the number of nodes in. the same
cluster that link
to 4 same-out: the number of nodes in the same clusteri. links:to; diff-in:
the number of nodes in
other clusters that link tóí; duff-out; - the number of nodes in other
clusters that i links to; same-in-
ratio: the proportionOf*Iinking nodes from the same cluster; same-out-ratio:
the proportion of
in-linking nodes from other dusters; w-same-in: same-in scores where value of
in-linking Wogs
is weighted by its centrality measure; w-diff-in: diff-in scores where value
of in-linking blogs is
weighted by its centrality measure;. and per-cluster influence scores; similar
scores (raw and
weighted) for in-links from, and out-links to, each cluster on the map.
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
191101 in another class of output, and with respect to across clusters
comparative frequency
measures of link, text, semantic and other node and link-level events, content
and features, the
partitioning of the network. into sets of source nodes may allow independent
and comparative
measures to be generated for any number of items associated 'with source
nodes. These may
include such items as: a) the set of target nodes K in IVI;.b.) any subset of
all target nodes, including
Those on user-generated lists;- c) any set of target objects, such as all tins
for videos on
YOUTUBEE", -or all object IBMs on user-created lists; d.) any other tins; c.)
any text. string
found in published material from source nodes; f.) any semantic, entities
found in published
material from source nodes; g.) any class of meta-data associated with source
nodes, such as tags,
location data, author demographics, and the like. For any item i in a set of
items associated With
source nodes, the -following examples of measures may be generated per each
cluster: 1.) total
count: number of Occurrences of item within the Cluster (multiple occurrences
per source node
counted);- 2) node count: number of nodes with item occurrence within cluster
(multiple
occurrences per source node count as 1).--3).itenifeltister frequency: total
count/i4 of nodes in the
cluster; 4.) node/cluster frequency: node. countl# of nodes in the cluster;
S.) standardized
itemieluster frequency: multiple approaches are possible, including z-Sc.ores,
and one approach is
to use standardized Pearson residuals, which control for both cluster' size
and item frequency
across clusters and items in the set; and 6) standardized node/Cluster
frequency: multiple
approaches are possible, including z-scores, and one approach is to use
standardized .Pearson
residuals, or Cluster Focus Index scores 122. The higher the CFI score for the
item, the greater
the degree of its disproportionate use by the cluster. A score of zero-
indicates that the cluster cites
the source at the same frequency as the network does on average. Other
detailed data may be
possible to obtain, such as the top nodes in each cluster, lists of all nodes
in the cluster, lists of
relevant Internet sites that each of the clusters link to (which enables
identifying target -outlinks
where a Message can be placed in order to reach spedific clusters), the
relative use of key terms
across the clusters (which enables developing specific messages to
contnimicate to eachcluster),
a hitcount (the taw number of times each (unlink and -term was found within
all the identified
nod*, source: node and/or cluster geography and demographics, sentiment, and
the like..
101-1.11 Ferrexaniple, differential frequency analysis can be done on meta-
data, such as tags, that
areaSSOefated with different attentive clusters to facilitate cluster
interpretation. In the example,
bylort.Og cluster focus scores 122 on the meta-data tags, interpretations of
what the clusters are
-About may be derived without any manual review. The meta-data associated with
the clusters may
be.used to facilitate interpretation of the meaning of the clusters. In an
example, the meta-data
may be language independent, such as GIS map data.
21
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
101121 in another class of output, and with tweet to visualizations of the
partitioned network
124, a social network diagram may be generated and used to display link, tekt,
semantic and other
node and link-level events, content and. features ("event data"), such as that
shown in Mel The
network map may be static or it may be the basis of an interactive interface
for user interaction
via software, software-as-a-service (SaaS), or the like. There may be two
components to this
process of visualization: .I..) creating a map of source nodes in a
dimensional space for viewing;
and 2.) use of colors, opacity and sizes of graphical elements to represent
clusters, nodes and event
data. With the dimensional mapping component, multiple approaches may be
possible. One
methodl may be to use 4.-**Physics model"-or "spring embedder" algorithm
suitable .for plotting
largattetWorkdiagrains. The Fruchternian-Reingold algorithm May be used pkit
nodes in two
or three dimensions. In these maps, every node is represented by a- dot, and
its position is
determined by link to, from, and among its meighbors. The size of the dot can
vary according to
network metrics, typically representing. the chosen measures of node.
centrality. The technique is
analogous to a locally-optimized multidimensional scaling algorithm. With the
component related
to use of colors, opacity and sizes of graphical elements to represent
clusters and event data, nodes
May be colored according to selected cluster partitions, to allow easy.
identification of various
partitions. This projection of the cluster solution onto the dimensional map
may facilitate intuitive
understanding of the "social geography" Odle online-network. This type of
visualization may be
referred to as a "proximity cluster" map, because proximity of nodes to one
another indicate
relationships of influence and interaction. -further, projection of event data
onto the Map may
enable powerful and. immediate insight into the network context of various
Online events, such as
the use of paiticular worth or phrases, linking to particular sources of
information, or the
embedding of particular videos. This may be produced as static images, and may
also be the basis
of software-based interactive tools for exploring content and link behavior
among network nodes.
101141 in anotherelass.of output, and with respect to aggregate cluster
metrics 128, metrics may
be Calculated .for partitions at the aggregate- level. Eventmetrics may
include raw counts, node
counts, frequencies (counuf# nodes in. duster), normalized and. standardized
scores, and the like.
Examples typically include 'values, such as: the proportion of blogsin a
cluster 'using a certain
phrase; the number of blogs in a cluster linking to a target website;. the
standardized Pearson
residual (representing deviation from expected values based on chance) of the
links to a target list
-of online videos; the per clOter "temperature" of an issue calculated from
an.array of weighted-
value relevance markers; and the like.
101141 As described above, any particular set of cluster solutions for target
nodes may be se1eete4
and used in combination with the set .of cluster solutions for source nodes in
order to generate
additional outputs. Visualizations produced may include: I.) two-mode network
diagram of
22
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
relationships between clusters of sources and targets*. treated as aggregate
nodes and with tie
strength corresponding to link density measures; and 2) Second-mode ('eo-
citation") network
diagram, in which targets are nodes, connected by ties representing the number
of sources citing
both of them, and colors corresponding to cluster solution -partitions.
Another output may be
macro measurement of link density. To reveal and measure large-scale patterns
in the distribution
of links from source .nodes to targets* the matrix :M may beeollapsed to
aggregate link measures
among clusters of sources and clusters of targets. A series ofSXT matrices may
be used, with S
as the set of source clusters ("attentive clusters") and T as the set of
clustered targets ("outlink
bundles"). Thew matrices may contain aggregated link measures, including:
counts (C); the
number of nodes in source cluster s linking to any member of target sett;
deitsitieS-(d): c divided
by the product of the number of rnembers.in s and the number of members in t;
and standard scams
(S): standardized measures of the deviation from random chance .for counts
across each cell.
Various standardized measures are possible, with standardized Pearson
residuals obtaining good
results. Any of these measures may be used as the basis of tie strength for
two-mode visualizations
described above.
NW] In an embodiment, a density matrix may be constructed between attentive
clusters and
outlink bundles. The attentive clusters.may be represented as row headers and
the outlink bundles
may be represented as column headers. The density matrix may allow users to
see patterns in
attention between certain sets of wehsites and certain bundles. The density
matrix may provide a
way to identify similar Media sources. Further, the density matrix may provide
information about
-attentive clusters that may be based on particular verticals.
01161 Flaying described the process for .attentive clustering, we now turn to
examples of
applications of the technique and various related analytical applications
thereof fin measuring
frequencies of links between attentive clusters and outlink .bundles, thus
enabling identification
and measurement of large-scale regularities in the distribution of attention
by online authors across
sources of information.
101171 in an embodiment, and referring to Fla 2, a social network map_ of the
English-language
blottosphere is depicted. The social network map graphically depicts the most
linked-to blogs in
the English language blogosphere. The size of the icons representing each
individual blog may
be representative of a network metric, such as the norther of inbound links to
the biog. This
visualization depicts the Output from a Method. for attentive clustering and
analysis which
identified-attentive clusters of linked-toblogs, wherein the attentive,
clusters included authors with
similar interests
101.18) Referring to FIG. 3, the method for attentive clustering and analysis
analyzes hloners'
patterns of linking to understand their interests. The visualization in FIG. 3
highlights liberal and
23
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
conservative U.S. bloggers, and British bloggers as attentive clusters. By
zooming in on the
visualization, 'subgroups such as conservatives focused on economics or
liberals focused on
defense may be identified from. among the attentive clusters depicted.
101191 Referring to FIG. 4, the method for attentive clustering and -analysis
enables building a
custom network map. In FIG. 4, the network map features attentive clusters of
bloggers attuned
to these topics: environmentaliSts, -feministsõ -]political bloggers, and
parents. Subgroups Within
each topic may be delineated by a different color, a different. icon shape,
and the like. For example,
within the parent bloggersõ icons representing the liberal parent bloggers may
be colored
differently than the traditional parent bloggers. Sutprising relationships may
be discovered among
groups of bloggers. For example, in FIG. S. two parent bloggers with very
different Social Values
are closer in the network than either is to political bloggers who share their
broader political views.
10120] Referring to FIG. 6, each attentive cluster may have its own core
concerns, viewpoints,
and opinion leaders. The method for attentive clustering and analysis enables
identification of
blogs that are considered bridge Wogs, such as the one Shown .circled, which
indicates that the
blog is popular among multiple attentive clusters. The method for attentive
clustering and analysis
enables identification of whose opinions matter, about what, and among what
groups.
10121.1 Referring to FIG. 7, the steps of attentive clustering and analysis
may include constructing
an online author networkõwherein constructing the onlineauthor network
includes selecting a set
of source nodes (S.); as-et:of out link targets (T) from at least one selected
type of hyperlink, and a
set of edges (E.) between S-and T defined by the at least one selected type or
types of hyperlink
from S toT during a speCifiedlittePeried 702; deriving a set of nodes, T, by
any combination of
a.) normalizing nodes in T. optionally too selected level of abstraction, b.)
using_ lists of target
nodes thr exclusion ("blacklists"),õand 04. using lists of target nodes for
inclusion. ("whitelists")
704;:transfOrming the onlinetruthornatwo* into a matrix of source nodes in S
linked to targets
in -r 708; and partitioning-the onlineatithorinetWork into at least one set Of
source nodes with a
Similar linking history to form an attentive cluster and at least one .set of
outlink targets with a
similar -citation profile to form an out-link bundle 71Ø The steps may
optionally include generating
_a graphical representation of attentive clusters and/or outlink bundles in
the network to enable
interpretation of network features and behavior and calculation of
comparativ.estatistical measures
across the attentive clusters and outtalk bundles 712, Wherein at least one
element oldie graphical
representation -depicts a measure of an extent Of a type of activity within
the network; and
optionally measuring frequencies of links between attentive clusters and
oudink bundles enabling
identification and-measurement of large-scale regularities. in, the
distribution of attention by-online
authors across sources of information 714. The element of the graphical
representation may use
at least one of size, thickness, color and pattern to depiet.itlYpeof
'activity. Attentive clusters may
24
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
be visually differeotiated in the graphical representation by at least one of
*: C0104- -a _shape,
shading, and a size. The size 1)1 the object representing the attentive
clusters in the graphical
representation may correlate with a metric. The nodes, targets, and edges may
be collected from
public and private sources of information. Constructing the matrix may include
applying at least
one threshold parameter from the group consistingef: maxnedes, targetmax,
nodernin, mrgetmin,
maxlinks, and linkmin. Constructing the matrix may include applying a minimum
threshOldfor
the number of included nodes that. must link to a target to qualify it for
inclusion in the matrix.
Constructing the matrix may hick& applying a minimum threshold for the number
of included
targets that must link to a node to qualify it for inclusion in.theuuttrix,,
,constructing the matrix
May include _using blacklists...to. exclude Particular nodes, and whitelistS
to fore: inchision _of
particular nodes. The matrix_ may be a graph matrix.
10122] By identifying and measuring the frequencies of links between attentive
clusters and
outlink bundles, all manner of information about the distribution of attention
by online authors
across sourcea of information may be obtained. Various examples of the sorts
of information.,
visualizations, applications, reports, APIS, widgets, tools, and the like that
are possible using the
methods described herein will be described. For example, two playlists for
YOUTIJBErm videos
may be identified, one that has fraction, with sub-duster A the other :with
sub-cluster B. In another
-example, two RSS feeds may be organized that supply a user with items that
have more attention
from sub-cluster A versus :sub-cluster B. In. another example, a valence graph
may be constructed
that. depicts words, phrases, links, ob*-ts, and the like that are preferred
by one sub-cluster over
-another sub-cluster; such valence graphs may use aggregated sets of eluSters
defined by users to
display dimensions of substantive interest, such as in. FIG. _It In yet_
another example, works
from authors who are:most relevant in a particular cluster may be displayed
and then published as
a widget, :which may be custom-based on a valence graph, -as a way of
monitoring an ongoing
stream of information from that cluster. Clusters may be -customizable: within
the widget, such as
via a dialog box, menu itemõ or the like. Further examples will be described
hereinbelow.
(012.3j A -user may be able to, optionally in real time through a user
interface, select a stream of
informatien based on looking at the environment., zoom in based on clustering,
figure out a. valid
emergent segmentation, and then set Op t.*.ionitQrs to Watch the flow of
events, Such as media
objects, text, key words/language, and,the like, in real time.
101.241 In an embodiment, differences in Word frequency use by attentive
clusters may be used to
differentiate and segment clusters. For example, the attentive dusters
"militant feminism" and
"feminist mom" may both frequently use terms associated with feminism in their
publications, but
additional use of terms related to militantism in one case and maternity in
another case may have
been used to subdivide a cluster of feminists into the two attentive dusters
"militant feminism"
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
and *100)iniataftom." In extending this concept, not just word usage but the
freqUency .of Word
litoge,-.thay..altoi be useful 41 segmenting clusters. For exaniple;,
iifilusters &Patera* the ones
actually -doing home sChooling did not use the term "bome school" frequently,
butaatber used the
taaan,horne education" with greater frequency. By identifying the specific
language/words used
by a cluster, the system may enable crafting messages, brands, language, and
the like for particular
clusters. In an embodiment, an application may automatically craft an
advertisement to be placed
at one or more outlinks in an outlink bundle using high frequency tents used
by an attentive
cluster. Further in the embodiment, the advertisement may be automatically
sent to the
appropriate ad space vendor for placement at the one or more outlink.s.
101.251 In an embodiment, a method of using attentive clustering based on
analysis of link
structures to steer a further data collection process is provided. The data
collection may include
collection of web-based data, such as, for example, clickstream data, data
about websites, photos,
emails, tweets, bloas, phone calls, online shopping behavior, and the like.
For example, tags may
be collected automatically or manually for every website that is a node. The
tags may be non-
hierarchical keywords or terms. These tags may help describe an item and may
also allow the
item to be found again by browsing or searching. In an example, tags may be
associated in third-
party collections such asliBLICIOLIS tags, and the like. In another example,
theft web crawlers
may extract meta keywords and tags included within node hard. Further,
specific keywords and
phrases may be exported to a database. in yet another example, the tags may be
generated by
human coders. Once a cluster partitioning exists, the system may do
differential frequency
analysis on the tags that are associated with different attention clusters. By
sorting cluster focus
index (CFI) scores along with the tags, the system can come up with an
interpretation, of the
meaning of a cluster without requiring further analysis of the cluster
itself:: twin embodiment, the
system may apply a further data collection. process in order to associate
respondents to a survey
and their riews.sources with various corners of the interact landscape. For
example, the influence
of a particular news outlet flerOSS a segmented environment of the online
network may be obtained
by examining clustering in conjunction with a downstream data collection
process, such as
obtaining survey research, elickstream data, extraction of textual features
for content analysis
including automated sentiment analysis, content coding Of a sample of nodes or
messages, or other
data.
101261 in an embodiment, clustering data may be overlaid on GIS maps, "human
terrain" maps,
asset data on a terrain, cyberterrain, and the like.
101271 In an embodiment of the present disclosareaa method of determining a
probability that a
user will be exposed to a media source given a known media source exposure is
provided. The
media source may include newspapers, magazines, radio stations, television
stations, and the like.
26
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
For example, a user Who may be exposed to a particular media source may be
clustered in a
specific -attentive cluster. .Accordingly, the system may. 'determine that
users in that particular
attentive cluster are more likely-to be exposed to. another media source
because the second media
source may also be present in an ondink bundle preferred by the cluster.
101281 In .an embodiment of the present disclospre,, a method of attentive
clustering on a meso
level is provided. The method may enable identifying emergent audiences -
(Attentive Clusters)
and monitor how messages (as specific as a single article in print; as broad
as core campaign
themes) traverse -cyberspace. The method may involve mapping the attentive-
clusters - where
messages have, or are. likely to find, receptive audiences. Mapping may enable
identifying opinion
leaders, -and information sources,online and offline, which help shape their
views.
101291 The method may enable identification of the mindset/social trends of a
group of users. For
-example, the system may be able to associate an attentive -cluster with a
known network, such as
pofitical party, a political movement, a group of activists, people organizing
demonstrations,
people planning protests, and the like. Via the ability to associate attentive
dusters with particular
groups of people, the system may, be able to track the evolution of a movement
or identity over
time. Further, if a cluster supports a political movement, the 4r.#610 may
track the 'impact of the
political movement of the cluster on society. The systemmaytrack if the
political movement has
been accepted by majority of the people of the society, rejected by the
society, if there is debate
about the political movement, and the like. Accordingly, the method may enable
growth of a
brand, sale of a product, conveying tirae0iige, prediction of what people care
about or do, and the
like.
101.301 -bran embodiment of the present disclosure, _a system and method for
multi-layer attentive
clustering may be provided. In the system and method, attentive clusters may
be tracked across
various layers of a social -segmentation such as specific social media
networks (Twitter",
.FacebookTm, OrkutTM. and the like), a blogosphere, and the like.. The system
may be able to track
development of an attentivetluSter asinglelayer or across multiple layers at
every stage of the
development of the cluster. When different layers of online media (such as
tiveblogs, microblogs,
and a. social network service) are clustered individually, measures of
association may be created
between clusters across layers, based on density .of byperlinka between them,
commOn identities
of underlying authors, Mutual. Citation of the same soureeSõ mutual preference
Ibt certain topics or
language, and the like; .The system may also traCk the major moot* or:Ottiaqs
at every: stage of
development. of the cluster.
10.131I For example, the growth of an attentive cluster supporting a.
political movement may be
tracked back in time and over a period of a time. In the example, once an
attentive cluster may be
identified, the system may examine the nodes associated with specific players
in the attentive
27
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
cluster in order to determine characteristics, such as. who is talking to
whom, identi.kkeyliOdes
or hubs that link, many other layers and/or media sources; .identify apparent
patterns of affinity or
antagonism among clusters or other known. networks, who may have started the.
political
movement, when the political movement may have started, what messages were
used at the
forefront of the political movement's establishment,, the size of the
movement, the number of
people Who initially joined the-political movement, growth of the political
movement, influential
people from various stages of the political movement, and the like.. In this
eXample, all of the
analysis may be 'confined to activity in a single layer of a social
segmentation or it may be
undertaken across multiple layers. Continuing with the example, the impact of
the political
movement- on society may be examined by tracking 'the penetration of an
attentive cluster or its
message across layers or the. expansion of the attentive cluster in a single
layer. 'Likewise, attentive
cluster -analysis may enable predictions. For example, an attentive cluster
may be tracked in a
single layer, such as by monitoring the number of Twittetru followers (or
other applicable social
'platforms), the frequency of new followers added, the content associated.
with that attentive
cluster, inter-cluster associations, and the like, to determine if a political
movement may be being
spawned, expanded, diminished, or the like. in an embodiment, the socio-
ideological
configuration of the people who spawned the political movement may be evident
from analyzing
one or more-of a hlog layer, asocial networking layer, a -traditional-media
layer, and the like.
101321 For exempt; a.Twittormi (or other applicable platform) map may be
formed where each
colored dot is an individual Twitter" m account and the position is a function
of the. "follows"
relationship. People are close to people they are following or who are
following them. The Pattern
of the map may be related to the structure of influence across the network.
101331 In an embodiment, the system-may be deployed on a social networking
site to identify and
track attentive clusters and linkage patterns associated with the attentive
clusters. For example,
the system for attentive clustering may be applied on FaCebookm to identify
attentive clusters in
the FaeehookTM audience and track the cluster's activity within FacebookTM In
an example, the
system may be used to identify a group of _people who may be susceptible to a
message. By
identifying and tacking an attentive cluster in the Facebookrm layer that may
be susceptible to a
message, downstream activities, such as organizing in response to the message,
may be examined.
For example, an attentive Cluster of university students May be presented with
a message regarding
a proposed law lowering the drinking age. The system may track activity Within
the cluster related
tothe message, identify new groups formed around the topic of the message,
invitations to other
groups. regarding the message,.opposition from other groups in response to,
the message, and the
like, Indeed, the system may be able to track the formation of new attentive
clusters in the
Fac.ebookTM layer in response to the message. In this case, the system may
identify individuals or
28
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
-groups that link to one another who share a common interest or target of
attention, such as
concerned parents opposing the proposed la*, anti-government groups supporting
the proposed
Law, child advocate groups opposing the law,. and the like. Discoveries
related to the original layer
may be applied to strongly associated clusters in other layers. For instanceõ
determination about
the interests of a cluster in the Facebookrm layer may he used to drive a
communications or
advertisinvarategy in associated clusters of other layers such weblogs or
Twitterm.
(0134j Measures for characterizing contagious phenomena propagating on
networks may include
peakedness, -commitment (such as by subsequent uses and time range), and
dispersion (including
normalized concentration and cohesion) and will belurtherdescribed herein.
101.351 In other embodiments,_ two-Mode networks may he generated by
projecting Modes one
-onto another. For example, certain social networks may not allow handling of
individual data, but
may allow public page data to be accessed. In this way, data from individuals
who comment on
public pages may be obtained. Public pages may be treated as a two-mode
network that is
collapsed to one mode, For example, a- two-mode network may be formed from two
classes of
actors, people and cocktail parties that the people attend. One class of
actors could be labeled I-
S and the other dtt=D; to generate a scatter diagram depicting a two-mode
network, either a network.
of cocktail parties attended by the same people- or a network. of people who
attended the same
cocktail parties.. Likewise, networks may be formed based on who participates
in the stream of
objects that come from different public pages, the relationship between public
pages such as if
there is a_ direct "like". relationship between public pages, weighted by how
many people
commented on objects from two or more pages, and the like.
j01.36I These.- data may be clustered .as described herein. hi embodiments,
the weight between
public pages indicated by the number of users commenting on object from both
pages may he
used to visually indicate a stronger connection between pages with higher
weights.
101371:PloSteringofthispublit page data may result in the formation of poles.
For example, twe
poles may font...Whore:one set of pages is interacted with by one population
and .another set of
pages interacted with by a very different population. There may be individuals
who are interacting
with both of these sets of pages at either pole. In any event, in the process
of attentive clustering,
users who are most tenuously .connected to anything are forced to the outer
edges of the cluster
map
101.381 In an embodiment of the present disclosure, a method of analyzing
attentive clusters over
time is provided. The analysis of these attentive clusters may enable the
system to depict changes
in the linking patterns of attentive clusters, over a time period. Further,
the analysis may allow
depiction of any changes-in the structure of the network itself.
29
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
101391 in an embodiment, a time-based reporting method may be Used by the
system to
demonstrate the effects of .events/actions throughout a network of attentive
clusters for a period
of time. in the method, bundles that marbelists of-semantic markers, including
text elements
embedded in a post. or tweet, links to pieces of online content, metadata.
tags, and. the like, may be
tracked in clusters across a network, such as a blogosphere.
101401 For example, a bundle of semantic triarkerS.related to obesity may be
tracked over time to
determine how the topic Of obesity is being discussed. in the example, a
particular bundle (with
text, link and meta data elements) can be tracked across clusters to set where
they are getting
attention or not. The measure of attention maybe defined as a "temperature,"
The "temperature"
is based conceptually on Fahrenheit temperatures (withotitnegatives)aS
compared to other issues
where 100 is very, hot and 0 is ice cold. The method may have a track iitg
report as an output for
tracking issues in a map across time. In this example, the tracking report may
be fOcused on a
collection of blogs most focused on childhood obesity organized into attentive
.clusters over a
moving. 12-month period of time. The blogs. may be clustered broadly into
policylpelitics, issue
focus, culture, fatnily/parenting, and food attentive clusters. There_ may be
sub-clusters defined
for each of those clusters, such as conservative, 'social conservative,
andliberal sub-clusters under
the policy/politics cluster. The report may indicate the issue intensity for
each cluster/sub-cluster
by assigning it an- average temperatureper blog of conversation on the broad
topic of childhood
obesity within, each group. The report may indicate the issue distribution tbr
each. eluster/sub-
cluster by calculating a percentage of childhood.obesity conversations taking
place on blogs. not
in the map and within each cluster Within the map. Continuing with this
example, specific terms
may he tracked across the chtstersisub-clusters over time and the method may
indicate an average
temperature based on the uses of specific terms in blogs within each cluster.
In the example, the
term "school lunch" has a high "temperature" in certain issue focus clusters,
liberal policy clusters,
and %odic clusters and steadily increased overthe last eight moving 12-month
periods. Similarly.
the intensity of sites, or the average temperature based on links to specific
web _sites on Wogs
within each cluster, maybe provided, by the report. The intensity of source
objects, or the average
temperature based on the links to specific web content (articles, videos,
etc.), may be provided by
the .report, The intensity of sub-issues, or the average temperature of
conversation on identified
issues -defined by a set of terms and links, may be provided by the report.
In. the 'report, specific
terms may be tracked on a monthly and per-Cluster basis, specific sites may be
tracked on a
monthly and per-cluster basis, and -specific Objects may be tracked on a
monthly and per-cluster
basis.
(01411 In an exemplary embodiment, the -system may identify and track
structural changes in a
network. For example, during the recent US elections, Wogs appeared
instantaneously that were
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
anti-Obama, Pro-Patin, or Pro-McCain but were outside the conservative
blogesphere. This rapid
change in the network structure. May be indicative Of A coordinated,
synchronized .campaign to
message and -biog.
101421 in an embodiment of the present disclosure, a method of attentive
ctusteriag by partitioning
an author network into a set of source nodes with similar adoption and use of
technology features
is provided. For ekample, instead of a website being a target of attention
foran attentive cluster
.or around which an attentive cluster forms, a feature or a piece
oftechnology, such as an embedded
Facehookim 'like" button, may be a target of attention or Clustering item.
101431 in an embodiment, a method of creating clusters of people and
describing probahilistie.
.relatinnShipS with other clusters, such As words,- brands, people, and the
like, is provided. The
system may describe any probability of any relation between them.
101441 To identify what an attentive cluster links to more than the network
average or what words
and phrases they use more than the network average, .a cluster focus index
score (CFI) may be
calculated. CH represents the degree to whieh, an event, characteristic, or
behavior
disproportionately occurs in a particular cluster, or a partieuiar cluster,
relative to the network,
preferentially manifests an :event, characteristic, or behavior. For example;.
CFT Score could be
generated for a particular cluster across a Set of target nodes, representing
The degree to which a
particular target. is disproportionately, and preferentially cited by members
of theparticularcluster,
or the degree to which the particular cluster, relative to the network._
preferentially cites the target,
The CFI gives a sense of what is important to an attentive cluster, where they
go fbr their
information, What words, phrases and issues they discuss, and the like. FIG. 9
depicts a graph of
cluster focus index spores. for targets: of a conservative-grassroots
attentive cluster. The targets
circled on FIG. 9 (F through .Y) are those that everyone in the- network links
to, according to their
CFI. The targets circled in 'FIG, 1.014.0trough E) are those that are
disproportionately linked to
by the conservative4rassroots attentive duster. according to their CFI,
101451 In an embodiment, a method of identifying .websites with high attention
from an identified
attentive cluster Or author is provided. The method may include determining
the websites
frequently or preferentially cited by identified authors by examining the
websites' cluster focus
index (CFI) score, Further,, the method may include automatically sending or
placing
advertisements, alerts, notifications, and the like to the websiteS. For
exampitia social network
analysis may generate a network map with thousands of nodes clustered into
attentive Clusters. In
an example with bloggers, influence data that results from the network
analysis may be influence
metrics for sites from across the Internet which blogsers link to, including
mainstream media,
niche media, Web 2.0, other bloggers, and the like. These are the influential
sources (also called
outlinks, or targets) used by specific groups of nodes across the map. For
example, influencing a
31
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
targeted cluster or bloggers can. often be accomplished by. targeting these
sources,'"upstream" in
the information cyele, rather than going after the bloggers directly. in Other
.embodiments,
influence data may be metrics that -reveal network influence among bloggers
directly. Sloggers
are usually thought of as simply being more: influential or less, but this
.data lets the analyst
discover which blogs are influential among which online clusters (segments), a
far more granular
and targeted approach. Each of these data sets can be sorted to
ex.amine_either influence over the
entire map or disproportionate influence over particular clusters (i.e., how
to reach particular
audiences), Cluster targeting can be further refined to identify which nodes
in A specific cluster
have influence on: any ofthe other clusters:orti.-the map. Because the
conversation within social
Media covers a wide variety of topics, =source and network influence alone do
not necessarily
reflect influence on a specific topic. A relevance index metric for discussion
regarding particular
topics, events, and the. like may be added to a social network analysis to
identify which nodes are
most focused on this topic.
101461 For both data sets there are two main sorts metries.:representing
influence. First are metrics
representing the. influence of nodes in the one-mode network. (set of source
nodes 5) as a whole,
Or directly among particular clusters or among specific other nodes. For
example, for any given
node in S, count (also called in-degree) is the number of other nodes in S
that link to it. Count
can be calculated across: the whole..map, or per cluster. Second, score can
be. calculated that shows
the influence of target nodes (nodes_ in T or T') on clusters of nodes in, S.
Count can also be used,
and Crl.:.scorea can be calculated that represent the influence of partictilar
targets on specific
attentive clusters. In other word, how specifically interesting or
authoritative the-target is for that
cluster. Relevance! index scores may - for nodes may also be calculated. using
lists of semantic
markers, to provide further metrics of value for targeting communications,
advertising, and the
like. Depending on the communications strategy, specific- sorts of the data
will create lists of
likely high-value targets for further action. While count, CFI. And relevance
index'-scores are all
important, they can be combined in order to maximize certain objectives.. The
following use case
examples include combining count and relevance into a targeting index, by
multiplying their
values. Other, more complicated maximization formulas are possible as well.
The examples
demonstrate specific influence. sorts that can be generated from the 'Russian
.network data tO
address each use case. The network data is based on the linking patterns of
the nodes in the.RuN-et
map over a nine-month 'period ending in February 2010.
101471 Use Case I and Use 'Case '2 involve finding influential sources.. Use
Case 1 involves
identifying sources with the.rnost influente.over the entire imipby doing
a.sort using the highest
values of count. While extremely influential, and in many cases suitable for
advertising
32
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
campaigns, these universally salient 'sites also tend to be much harder to
reach out to than sites
that are smaller-but- specifically important-to-targeted segments.
101481 Use Case 2 involves identifying sources that reach a targetedclu.ster
by sorting on sources
by Cluster Focus Index- CFIs may be sorted for any of the attentive cluster.
Count metrics from
the map as a whole and from the targeted cluster can be. used to further
prioritize for action. This
sort is the equivalent: of identifying traditional media trade press, the go-
eo :siteS for the -selected
Segment. Frequently, these include specifically influential bloggen in
addition to niche media
and other sources.
101491 Use Cases 3-6 involve finding influential nodes. Use Case 3 involves
identifying the
greatest network influence by sorting the nodes by indeg (total: number of
links from other nodes
within the entire network). This sort specifically identities the
networit's:"Aslist" nodes, the most
influential network members (bloggers). Like prominent sources, these are
often more difficult
to. reach than more targeted niche influentials, but they contribute greatly
to. spreading viral niche
.messages across. the wider network.
101501 Use-Case 4 involvesfinding the most targeted influencers for a
particular cluster by sorting
the Cluster Focus Index scOres for a targeted cluster to find nodes with
cluster-Specific influence.
This :identifies- the nodes with particular influence, interest, or prestige
.arrione the target cluster.
These nodes tend to be much more "on topic" than others, and much easier to
reach that map-wide
A-list nodes. Cluster-specific influentials are not always from the target
cluster itself, which can
he very useful. for trying to move discussion between particular clusters.
Link metrics provide
further assistance in deciding turgefingprioeities.
101511 Use Case 5 involves following a particular topicat the map level by
sorting using topic
focus target scores, which combine links (network influence) and topic fOcus
index (issue
relevance). Formulas for calculating focus target score can be varied, but the
default may be to
multiply links by topic focus index.. This. may allow identification Of those
nodes'in the entire
map that discuss the target issue most frequently. These may be monitored 'to
gauge dominant
threads of discussion and opinion about the issue, and targeted for outreach..
101521 Use Case 6 involves targeting a particular cluster's conversation on a
topic by sorting
within a cluster by the topic focus target score. This may allow members of
the target cluster who
write about the target issue to be identified for monitoring Or persuasion:
Variations of the formula
for combining influence and relevance metrics into a single targeting metric
can be used to bias
the sort toward relevance, Or tOward influerice, depending on strategic
objective.
10.1531 In an embodiment, a proximity cluster map method may be used-to
visualize 1.24 attentive
cluster-based data and generate a network map. in the method, attentive
clusters and their
constituent. nodes may be displayed in a proximity cluster map. Nodes in the
network map may
33
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
be represented by individual dots, optionally represented by different colors,
whose size is
determined based on the numberof othernodes on the map link to-them. A general
force may act
to move dots toward the circular border of the map, while a. specific force
pulls together every pair
of nodes connected by a link. In static images or an interactive visualization
via software
connected to a database, nodes may receive a visual treatment totlisplay
additional data of interest,
.For-example, dots representing nodes may be lit or highlighted to represent
all nodes linking to a
particular target, or using a particular word, with other nodes darkened. In
another example dot
size may be varied to indicate aselected node metric.
(01.541 In an embodiment, avalence graph method may be
use4.t0..vkataiiie124.attentive cluster-
based data and generate tt. valence graph. In the method, targets Of attention
or Se-Mantic elements
occurring. in the output of nodes may be displayed in a valence graph. The
valence graph method
may be understood via description of how a particular valence graph is built,
such as a Political
Video Barometer valence. graph (FIG. 8) useful for discovering what videos
liberal and
conservative bloggers- are writing-about. This particular valence graph may be
used to watch and
-track videos linked-to by bloggers who share a user's political opinionsõ---
view clips poptdar with
the user's political enemies,"and the like,
101551 The videos shown in-the Bammeter are Chosen by queries againstt, a
large database built by
network analysis engines .performing network selection In- Periodically, a
crawler (or "spider")
visits millions of Wogs and collects their contents and links. Next, the
system mines the links in
these blogs to perform partitioning 104 and forms attentive Clusters based on
how the Wogs link
to one-another (primarily via their blog rolls), and, over time, what else the
bloggers link to in
common. Attentive clusters may be large or small, and. the bigger ones. can
contain many sub;
clusters and even sub-sub-clusters. In embodiments, determining what the blogs
have in common
may be done by examining mewdata, tags,. language analysis, link target -
patterns, contextual
understanding technology, 'or by human examination of the blogs = or a subset
thereof, in the
example, American liberal bloggers and American conservative bloggers form the
two largest sets
of clusters in the English language blogosphere, and the Barometer draws upon
roughly the 8,00(1
"most linked-to" blogs- in each of these groups to position the videos on the
graph by calculating
proportionsof links to each target by the two political cluster groupings.
101561 The riatOttetertnay be Continually updated by scanning the blogs
periodically, looking for
new links to videos (or videos embedded right in. the Wogs). By counting,
these links, it can be
determined what videos political bloggers- are promoting. In embodiments, the
link eount may be
-displayed on the valence graph using an identifier such. as icon or marker.
In this example. some
videos are linked to almost exclusively by liberal bloggers, some are linked
to mostly by
conservative bloggers.. and a few are linked to more or less evenly by both
groups. Once the
34
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
system determines that a video has traction in the political clusters, it
scans through data from
other parts of the hlogosphereto count how many "non-political" bloggers link
to it as well.
j01.5-71 The Political Video Barometer example illustrates one. kind of
valence graph and the
insight that can be gained and. the applications that can be built based on
the method and the data
obtained by the method. It should be.understood that the method may be used to
examine any sort
of potentially cluster-able data, such asuchriology, celebrity gossip, the
.use of linguistic elements;
the identification of new sub-clusters of particular interest, and the like.
All aspects of the valence
graph methodõand the underlying attentive clustering analysis, may be
customized along multiple
variables to enable planning and monitoring campaigns ofall kinds.
101.581 man embodinfent, a multi-cluster recut comparison Method: may enable
comparing cluster
focus. index (CFI). scores of multiple attentive clusters. The CFI score:may
be a measure of the
degree to which a particular outlink is. of disproportionate interest to the
attentive cluster being
analyzed; in other words, the CFfindicates what link targets are of specific
interest to a particular
cluster beyond. their general intereSt: to the network as a Whole. In an
example, X may be.theCFI
score .for cluster A and Y:may be tlie(71 score for duster B. The multi-
cluster focus comparison
Method may compare the two clusterS, A and B, based_ on their CFI scores, X
and Y. This would
allow a user to discern elements of tonlitTfoll interest vs. divergent
interest between the two
clusters. Insights derived from this method would he of great value in
creating and targeting
advertising and communications campaigns.
101591 in another embodiment, link targets, semantic events, and node-
associated metadata may
be scattered in an x-y coordinate space, and the dimensions of the graph may
be custom-defined
using sets -of .clusters grouped to represent substantive - dimensions of
.interest. for a particular
analysis. Elements are plotted on X and Y according to the proportions of
links from defined
cluster groupings. For example, and referring to FIG.. 11, using data from the
.Russian
blogosphere, the top 2000 link targets for Russian bloggers may be plotted
such that the proportion
of links from "news-attentive" 'Wog clusters vs. links from "non-news
attentive" clusters
determined the position on Y, while the proportion of links from the
"Democratic Opposition"
cluster vs. the "Nationalist" Cluster determines the position on X, as shown
in FIG. II . In another
example, popular outlink targets for the US blogosphere May be displayed With
the X dimension
representing the proportion Liberal vs. Conservative bloggers linking to them,
and the proportion
of political bloggers of any type vs. non-political Wagers represented by the
Y dimension, as
shown in 'FIG. 13. Various data may be visualized in the graph .associated
with the dusters of
news,attentive and political bloggers,_ such as meta-data tags, words, links,
tweets, words that
occur within 10 words of a target word, and the like. These visualizations may
be used in
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
interactive software allowing user-driven exploration of the data. graphed in
valence space,
optionally allowing user-defined sets of dusters to be used in Calculating
valence. metrics.
101601 Irt an embodiment, a. method of node. selection 1-10 based on node
relevance to a defined
issue, also known as semantic slicing, is .provided. Semantic slicing may
involve .clustering
according to a relevance bundle. A relevance bundle may include one or more of
key markers,
what the nodes may have linked to, what the nodes have posted, text elements,
links, tags, and the
like. In essence,, semantic slicing involves pre-screened nodes for relevance
based on semantic
analysis.
10.101.1 iTherelevarice buudieernay.tv used tO sortibrough all of the network
data to select the top
:high relevance nodes. In an embodiment, ii=ciiitnin-inapning ofiti:Wb-SOf
Odle link economy may
bedcine.
101621 In an embodiment, semantic slicing may enable generating a
contextualized report of
interest to a user on an industry level. Semantic slicing may enable focusing
attentive clustering
on :selected vertical markets. The vertical markets may be a group of similar
businesses and
customers who may engage in trade based on specific and specialized needs.
Lists of semantic
market* such as key. words and phrases, links to relevant webSites and online
content, and relevant
metadata tags, are built Which represent the relevant vertical market
Relevance metrics are
calculated forcandidate nodes, and a selection, of high-relevance nodes..are
mapped and clustered,
Continuing. the example, the semantic slice may be done to analyze an energy
policy vertical
market by focusing the attentive clustering around one or more selected,
highly relevant nodes.
Thus, the attentive clusters may be more specific to identified domain
interest of interest or vertical
market. In this example, instead of just forming an _attentive cluster of
Conservative. bloggers, by
focusing attentive clustering on one or more key markers related to energy
policy, the attentive
clusters discovered include topic-relevant segmentations of particular kinds
of Conservative
bloggers discussing the issue, such as Conservative-Grassroots and
Conservative-Beltway.
Additional high-relevance attentive clusters may be identified, such as
Climate Skeptics. :Middle
East policy, and the like. Cluster focus index scores may be used to determine
whatsites everyone
in each cluster links to and which sites are preferred by the cluster. In an
embodiment, semantic
slicing may be done using a singlenode, such as a particular website, a
particular piece of eontent,
and the like. In. an embodiment, semantic -slicing may be done over a period
of time to enable
monitoring the impact of a campaign.
101631 man embodiment, a tool, such as software-as-a-service, for enabling
users to define one
or more semantic bundles for attentive, clustering and as the basis of report
outputs is provided.
The tool may be an on-demand tool that may be used for semantic slicing._ In
such models, a user
may declare a semantic bundle of nodes. and/or links prior to attentive
clustering.
36
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
101641 in an embodiment, the system may provide an application programming
interface (API)
ibr delivering a segmentation to track one ormore particular chiSters of
attention, or track how An
audience is interacting with. a piece of content and the like. 'Me data about
the various clusters
may be collected directly. from the API. For example, a user may wish to track
a cluster. The user
may enter keywords related to the duster in.a search option provided by the
API. Thereafter, the
tool may track. various websites and report hack. the weblinksiand data "that
may be relevant: to the
cluster. The API may be used to interact with a valence graph at various
resolutions. The API
may provide segmentation data and metadata derived, from the segmentation to
other analytics and
web data tracking firms, .for use in. their,own. client-facing tools and
products. The segmentation
and resultant data from attentive clustering provide an additional dimension
of high value against
which third-party tools and. other analytic capabilities such as automated
sentiment monitoring
may be leveraged.
01651 In an embodiment, the system may enable real-time selection of elements
to visualize
based on attentive clustering of .social*edia. 'The system may facilitate
selection of a stream of
information based on looking at the environment, zooming in on a data. element
based on
clustering, determining a valid emergent segmentation, and monitoring the flow
of events in real
time. The events may include media objects, text, 'key words/language, and the
like. For example,
the real-time selection of elements. may facilitate an analysis, of
trends/events .especially for
financial purposes.
101661 In an embodiment, a search_ engine may be provided that prioritizes
search results being
displayed tO a user based on a determination of real-time attention including
attention from a
particular cluster or set of clusters. A user may be able to customize the
prioritization of search
results, such as by getting. real-time attention from a particular cluster,
from a particular sub-
cluster, and the like.
101671 In an embodiment, a search engine is provided that searehes within only
those
Sites/accounts with high cluster focus for a chosen segment. For example, a
(300GLETM search
may be restricted to the .30 websites with the highest CFI scores for the Dirt
Bike racing cluster of
OAKLEY's IWIITERT-m, followers map. Thus, the search may only return results
from a. list of
key influential sites related to the chosen segment. In other embodiments, the
search may be
restricted to websites (or domains within them), with a particular CFLSt Ore,
Websites (or domains)
that meet a threshold CFI score, websites that fall into a range Of CFI
:Stores for a Chosen sement
websites with a particularM score, .and The like. In an embodiment, the
search query-may restrict
the search to particular 'websites that are identified based on the. CFI
scores. In an embodiment,
the search query- may be restricted by CFI score of a website and the CFI
score restriction may he
indicated in the settings of the search engine. In other embodiments, the 'CFI
score for sites to
37
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
starch may be indicated in the search string itself,: for example, a user may
indicate -a particular
search they want to perform and they maybe provided with a slider bar where
the user indicates
that the search should be restricted to those. websites with a CFI score
falling into the range
selected on the slider bar. The slidermay be provided with a normalized scale,
such as ascribing
I to low CFI scores and 10 to high CFI scores, such as using a linear,
logarithmic, or other scaling
process. The system may then search a dittabaSe ofWebsiteS.for the range of
CFI. scores to identify
one or more websites to which to limit the search. These websites are then
included in a starch
string that is provided to a search engine.
101681 Similarly, the search can be-restrktedloonly.speeitic.conterit, or
s.pecificeontent may be
promoted to high ranking within a search, leaving other Contentio.the kiiver
ranked .rtõ*-itilts. One
way to do this restriction is to utilize the valence mapping functionality of
the system. As
described herein, a valence graph may be constructed for a Chosen segment.
that depicts words,
phrases, links, objects, and the like that are preferred by one cluster over
another cluster, content
indicated in the valence graph may be indexed by the system and only that
content in the valence
:graph may be searched by a search engine. .Further restriction of the content
may be employed,
such as by websiteõ enseore, and the like.
101.691 In an embodiment, attentive clustering and related analyses may result
in identifying
issues, attitudes and messaging language that. may be specific toõdisco.urse
for a target market, and
may be suitable II-yr presentation in a report. For example, in a clustering
of bloggers sympathetic
to Arts in Schools, by examining intra-cluster linking patterns, it may be
determined that most of
the bloggers within each cluster tend to keep the discussion, within their
cluster except for the
bloggers in the "Interesting/teachers./educators"- cluster who have a tendency
to spread
conversation to each of the other clusters. This behavior points to an
opportunity to work with
these bloggers to spread messages across the space. In continuing with the
example, by examining
clustering relate.d to specific keywerds, websittaõ oudinks, objectS, and the
like, it may be
determined that: there is a broader discussion about education. and education
reform than about arts
and arts education. Therefore, a conclusion may be that introducing an arts
education message to
education discussions has more potential than introducing, arts education
messages to arts
discussions. In the report, various valence graphs may be presented, such as
cluster specific term
valence maps, maps of sources, OUtlink Maps, term sped-fie maps, issue Maps,
and the like.
Alternatively, the report may presented as a spreadsheet of data.
101701 in an embodiment of the present: disclosure, the report may feed into a
method of
generating a -campaign blueprint for both social and upstream media sources
and a method of
identifying influence inter-cluster and intnt-cluster in order to plan a
campaign. The blueprint
may include target audience, demographic details, objectives of the campaign,
flow of the
38
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
campaign, messaging to use in the campaign, otalinks to target, and the like.
Systems and methods
for measuring the success of a campaign in various online Segments and
generating targeted data
sets identifying sub-clusters specific to a uses identity or objective are
provided.
101711 In an exemplary embodiment, the campaign tracker may track data from a
variety of
sources to provide closed-loop return on investment (ROI) analysis. The tool
may parse the
information of each website accessed by the users, keywords entered, any
information about the
campaign, and the like. Further, the tool may track how people react to the
campaigns and which
ones are most successful. The campaign tracker may track and analyze results
in real-time to
determine the effectiveness of the campaigns.
101721 In addition, the tool may enable the system to generate reports for
clients: The reports may
include details about the campaigns such as campaign type, number of people
who have viewed
the campaign, any feedback from the people, and the like.
101731 In an embodiment, analyst coding tools (ACT) and a survey integrator
may support
distributed metadata collection for qualitative analysis to best, interpret
quantitative findings. The
tools may include an interactive visual interface .tbr navigating complex data
sets and harvesting
content. This interface may contain an interactive proximity cluster map which
can display
specific node. data, metadata, search results, and the like. This proximity
cluster map interface
may enable the user to click on nodes to see nodespecific metadata and to open
the node URI, in
a browser window or external browser. Using the tools, a user can add metadata
and view
metadata about any given blogger on a map. The tools enable grabbing whole
sets of blogs or
items to add to semantic lists, and may enable a user to define surveys so a
team of human coders
can open the website and fill out surveys.
101741 In an embodiment of the present disclosure, a dashboard may be
provided. The dashboard
may combine advanced network and text analysis, real-time updates, team-based
data collection
and management, and the like. In the. einbodiment, the dashboard may also
include flexible tools
and interfaces for both "big picture" views and minute-by-minute updates on
messages as they
move through networks. Using the dashboard, a user may define bundles and
track them in the
aggregate through networks over time. Using the dashboard, a user may be able
to see how
specific media objects are doing with a particular cluster over time.
101.751 In an embodiment, the dashboard may provide a burstmap feature
inAkrhic,b the history of
selected events or sets of events over a timeframe may be displayed. using a
proximity cluster map.
During playback, nodes in the map will light up at a time corresponding to
their participation in
the selected event or events. For example, at a time in playback representing
a certain date, every
node which linked to a particular YOLITUBErm video will light up, allowing the
user to see the
pattern of linking as it unfolded over time. Optionally, this burstmap feature
may include a
39
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
timeline view displaying event-related metrics_ over time, such at the number
of nodes linking to
a particular video. Optionally, the burstmap feature may include = EMS of
events- available for
display. Art example of a burstmap interface is found. in FIG. 1.2,
101761 In an embodiment, 'techniques disclosed herein may beused to_generate
social media maps
that visualize social media relationship data and enable utilization of a
suite of metrics on the data.
'Social media maps may be Constructed via clustering of various social media
communities
including. TW1TTERTm. FACEBOOKTm, blogs, online social media, and others. In
one
embodiment, the. clustering technique used may be manual, relationship-based,
attentive
clustering such as previously diSelosed herein, network segmentation, or
another, analogous
technique. The social media maps may be organized in portfolios that are
targeted to Market
segments or relate to an issue/topic campaign. Social media maps may be
offered Via an API or
as taw data to plug into a third party dashboard. Services related to the
social media maps that
may be offered include robust tools for searching, comparitte and generating
integrated reports
across multiple maps. searchable indexing and map browsing. Pricing for social
mediamaps may
be Via subscription, for one or more maps, a portfolio of maps, the whole
portfolio of maps, the
whole portfolio maps save some exclusive/custom items, or the like. Systems
and methods for
how to generate, utilize, update and offer social media maps will be further
described herein.
Mr] A comprehensive catalog.of social media maps and network segmentations may
be offered
and updated on a. regular basis. The catalog may include targeted portfolios
for key markets, such
as consumer goods, media and entertainment, politics and public- policy,
energy, science and
technology, government, and More. The catalog may contain maps for each layer
of the social
media system, such as: blogs. Twitter, social network services, forums, and.
the like. It may
contain maps for all major languages, countries and regions of the world..
Social media map data
may be used within partner dashboard systems, so that a range or commercial
tools can be
leveraged -by subscribers and so that the'social media Map data are "portable"
across various tools.
_In addition, a suiteof reporting tools may be used in conjunction with the
social, media maps.
101781 In an embodiment, one or more social media maps and network
segmentations may be
constructed via 'clustering of data from at least one social media community.
The social media
map or network segmentation may be offered via an API or. as raw data. The
social media
community May: be based on at least: one of a social media layer; a language,
a country, a region,
-or the like. In some embodiments, the clustering technique may be attentive
clustering, as
described previously herein, relationship-based, manual, network segmentation,
or the like.
Referring now to FIG. 14, relationship-based clustering of data from at least
one social media
community 1402 is used to construct one or more social media maps and network
segmentations
using the clustering 1404. One or more social media maps and network
segmentations may be
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
offered via an API 1408 or as raw data 1410. A report may demonstrate the
interaction of
nodes/links between the maps .1412.
101791 in embodiments, the maps may be generated by an autonomous process. The
autonomous
process may create maps based on one or more criteria, a. scope definition,.
an instruction, or the
like. For example, a social graph may be generated based on followers of an
individual or entity
in a social network. In another example, the map criteria may he semantically
based, such as
based on key words or hashtags. In yet. another example, the maps may be geo-
based, such as
based on which users/nodes are in a territory. In still another example, the
maps may he based on
previons.mappings. In this example, segments in other maps on health and
fitness may be used
*triangulate or iterate to a mapping of a new. category. In another example,
the map may be
based on an arbitrary set of accounts generated by a third party. One scenario
might be a mapping
of the social network accounts for all the users of a mobile application. In
still. another example,
the maps may be based on a nomination of individuals based on some criteria,
such as
demographics. Once generated, the maps may be stored and indexed.
101.801 In embodiments, maps may be based on CFI scores for dynamic data
(e.g.. YOUTUBETm
videos). -However, the amount of data may be increased to obtain a better
indication of what the
segment is communicating about whether data cart be obtained.on. the
influencers of a seement,
which may be coming from off the map. In addition to looking at data coming
from the segment.,
the system may be able to .access data from social media accounts that have
high CFI for that
segment (not just. the ones that are "in" the segment). Thus, calculating
cluster focus tbr the
dynamic data may be improved-. CFI scores may be calculated for a first
segment. Then, CFI
scores may be calculated for those influencers on the first segment. For
example, the first segment
may be followers of a particular art gallery but the system can also examine
the CFI fbr the first
segment's influencers, which may be several well-known Art Gallery aficionados
who may or
may not be followers of the particular art gallery. In embodiments, certain
maps may. be 'based
only on the CFI scores calculated for the influencers:
101811 A searchable index fora catalog of social media maps may be constructed
1414. Further,
social media maps in the catalog may be searchable. For example, the maps may
be searchable
by a keyword, a UK:, a semantic market, and the like. In embodiments, the
social media maps
may he indexed by oneor more Of a keyword, URI, or semantic marker so as to
form a searchable
index of social media maps. In embodiments, the searchable index. may include
metrics to indicate
a statistic regarding the social media maps. For example, thestatistic may
represent a dimension
of popularity, relevance, semantic density, or similar feature. For example a
search engine may
be enabled to return maps in terms of relevance by using certain statistics in
the searchable index.
41
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
191821 For example, a semantic marker may include it keyword, a phrase, AUR1,-
(tiode or object
level), a tag (such as those from book/narking and annotation services, meta
keywords extracted
from. :HTML, tags assigned by coders, etc.), and the like. Semantic markers
may also include
those used in particular social network environments, such .as TWITThRTm, and
may include
follows relationships, mentions, retweets, replies, hashtags, URIAargeta, and
.the like. .Any of
thew semantic mark.ers may be used to index a social Media map.
(0183j Based. on at least one of the search terms or the search results, a new
social media map
subscription may be suggested. For example, if a user searches a social media
map index for the
terms "Nissan LEAFrm,r:.'"eleetric-yetticle,' and leafstations.com,
subscriptions to social media
Maps such as autOrraibileS, eco-friendly products, and California trends may
be-stiggeSted..
101841 In an embodiment, a dashboard may be used for browsing, visualizing,
manipulating, and
calculating metrics for one or more social media maps constructed via
clustering of data from at
least one social media community. Clustering techniques may include
relationship-based, manual,
attentive clustering, or the like. . In some embodiments, the dashboard. may
be a third party
dashboard that supports visualization of data from clustering, wherein the
data may be delivered
by a raw data feed, an API plug-in, or any other data delivery Method. in
embodiments, the data
from clustering may be joined with or otherwise: inte.grated=with data from
other data sources to
fonn a new data set. The new data set may he similarly browsed, visualized,
manipulated, and
processed by dashboards.
101851 In an embodiment, APIs, dashboards, and partner tools may be used with
social media
maps for planning/assessment. For example, social media maps may be used for
enterprise
resource planning, business insight, marketing, search engine optimization,
intelligence, politics,
industry verticals, financial industry, and the like. For example, an
entertainment promotion
company may own a plurality of social media accounts. If they could navigate
sector-level
mappings related to gertreaW Music, they could use the maps to target music
genre-specific
messages using the most appropriate of those accounts for maximum
effectiveness.
101861 In embodiments, custom maps may be derived from mashing upsets of
social media maps.
101871 In an embodiment, the = social media maps may be constructed via
clustering (e.g.,
relationship-basedõ manual, attentive, etc.) of data from at least one social
media community
targeted to a specific market segment. For (*ample, the market segments may
include government
intelligence, public diplomacy, social media. landscapes in other countries,
pharmaceuticals,
medical, health care, sports, parenting, consumer products, energy, and the
like. In these
embodiments, themarket segment may be used. to index the. social. media maps.
101881 in an embodiment, a reporting product. may leverage social media maps
to demonstrate the
-interaction of nodes and/or links between social media maps. For example, a
multi-map report
42
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
may be generated comparing the. nodes and links in different social media
communities in. a
particular market/environment. The. reporting-product- may be integrated with
a dashboard or
analyties platform.: Multi-map reports generated by the reporting product. may
be used to
demonstrate various phenomena, such as .how particular items can he found in.
particular social
media layers. For example, a multi-map .report :may demonstrate how wehlog
hosts are having
customers driven .to them from TWTITERTm. In another eXample, a mufti-map
report may
demonstrate how FACEBOOMm pages are getting attention from a segment of
TWEITERim.
101891 In an embodiment, information derived from the social media maps,
including portions of
or the entire map itself, may be published or displayed as a map widget, which
may enable
monitoring an Ongoing Strewn of -information from one or more clusters or one
or more maps
information bring displayed that is derived from the social media map may be
customizable within
the widget, such US via a dialog box, menu item, or the like. A user may be
able to, optionally in
real time through a user interface, select a stream .of information based on
looking at the
environment, zoom in based on clustering, figure_ out a valid emergent
segmentation, and then set
up monitors to watch the flow of events, such as media objects, text,.key
words/language, and the
like, in mai time. The published, Widgetized map acts as a sensor network to
obtain a host of
behavioral data and leads that can be leveraged by the map's user or hosts. In
embodiments, users
may interact_ with other users' map widgets to discover content and
individuals/entities. Using
other users' map widgets, users. may grow their own, networks by engaging with
the content and
people/entities in the widget such as to start following a person or to
retweet an item.
(01901 There are at least- three- processes that yield attributes of nodes,
including calculating 4
relevance score, performing a CH bias weighting, and identifying nodes as
"allowed" or "not
allowed" (e.g., blacklist/whitelist). Automated social media map_ refresh may
leverage one or
more of these processes.
(0191] in an embodimeat and. referring to FIG. 15, a soeial ..media: map May
be automatically
refreshed via calculating -a relevance, score for nodes or bundles in the map
1502. and re-
constructing the map based on a relevance ranking revealed by the relevance
score 1504.
Semantic/relevance marker bundles may include lists of semantic markers like
key words, phrases,
relevant link targets, .accounts that are followed on TWITTERTm, and the like.
Semantic markers
may be manually curate& In an embodiment, the refresh process may involve
performing the
relevance search/semantic Slice that generated the original map for new
relevance/semantic
markers. A relevance 'calculation may be performed on the nodes to calculate a
relevance -score.
101921 In another embodiment, a social media map may be automatically
refreshed via positively
or negatively weighting at least one cluster based on a CH score calculation
1508_ and re-
constructing the map to modify the nodes in the clusters 1510. Modifying the
nodes may be done
43
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
to include positively weighted nodes and exclude negatively weighted nodes, PI
Scores for
clusters may be leveraged to evolve a-map in avertain direction. Clusters in
the map thatinclude
preferred/wanted nodes/links are -positively weighted. clusters are negatively
weighted in they
are deemed to not be relevant. Applyingweightiags to the map may enable
pulling in additional
nodes that are more relevant:. Weighting map clusters for the CFI bias
operation may be done by
humans.
(0193j in an embodiment, a social media map may be automatically refreshed
via, filtering out
unwanted nodes 1512. Inan.-enibodiment, a social media map may 'be
automatically refreshed via
obligatorily including nodes-that were not clusteredln-the original map 1514.
Semantic markers
that are known to not fit bated on their relevance ranking or for some other
reason are not allowed
are filtered out. in embodiments, nodes may be forced into the map whether or
not they were
identified in the relevance search/semantic slice. Curating black lists of
nodes may be done by
humans.
101941 In an embodiment, a. social media map may be automatically refreshed
via crowd-sourced
'Information regarding nodes and/or links That drive nodes to bundles 1518. In
an embodiment, a
'social media Map may be automatically refreshed via processing social media
map usage data for
trendslindieators 1510. Usage data may relate to one or more of what is
ignored, what it further
-explored, what is used, how clusters are grouped, what name/label is assigned
to a Cluster, what
color is used for a cluster, what order/position the cluster is placed in a
report and the like. Nodes
preferentially interacted with may be weighted moteheavily.
(01951 In embodiments, community feedback may influence each of the three
streams of
automated map refresh described herein. Community feedback provides an
indication of news,
events, inibrmation, etc. that may drive:addition of nodes to the bundles,
such as fOr example, if a
new website is a target link. This sort of feedback may provide feedback or
guidance as to the
CFI bias operation. For example,. if feedback suggests that a oluSter is
relevant, then that Clutter
may be positively weighted.
10196) Feedback and updating may be based on how people are using the maps,
such as,
understanding what they ignore, what they drill down on, what they use, how
they want to group
things, what name/label they assign a cluster, what color they use for a
cluster What clusters are
Most important to a client based on an order/position the client plates it in
a report, and the like.
Refreshing the maps-may leverage this captured information.
101971 In an embodiment, feedback = may be received passively from
clickabielinteractive maps
via a built-in feedback system. This feedback system may be. used as a
naiveweighting system.
In an embodiment, the map may include a flag available to provide commentary
or feedback.
44
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
191981 in. at.:0.*TiOle, a map may include raw clusters and human-made
groupings and the
attachment-a-other Soma metadata such as the coloring of a duster. The example
may be that
of the RussianbiogosPhere, Which may contain 40 clusters and 7-8 groups,
including 5 right wing
'Russlaknationalist groups and a liberal opposition group. Clusters may be
processed by hamarr-
-*signed re-aggregation, and metrics may be run against them to progressively
refine the clusters.
Different clients, even on a base map, May want to group things differently,
name a cluster in an
interface differently, color a cluster in an interface differently, and the
like. Users need to be able
to define groups, re-label clusters, select clusters and the like. Community
feedback may provide
observations as to how users are grouping the same map and that yields data
about which clusters
are related to each other that is "crow&Sourced" to the user. Users may define
the order in Which
the data are presented in the reporting, For example, a user may want to place
data on preferred
clusters higher in a chart. Cluster :ordering and positioning information is
customizable, which
can be harvested as an impOrtanee:Weighting by the community.
101991 in another example, Map users may contribute to map metadata to
generate a community
data set established and/or expanded by users.. For example, users could input
the gender of a
Tweeteriblogger. The user community itself may be a segmentable population.
The user
rommuaity can contribute to scraping a map for a particular topic. For
example, something about
a disease might appear in .various places: Consumer segments, Politics,
Medicaliseienee Sports,
and the like. User feedback may also help scope the size of the map. For
example, aatser may
ask.: Should the map be constructed on the first 5,000 targets or should
20,000 targets be used? In
an eMbOdiment, user-contributed data may be used to provide metadata for a
social media map
constructed via clustering (e.g., relationship-based, manual, attentive, or
the like) ofdata from at
least one social media community.
102001 In an embodiment and referring to FIG. 16, data, including user-
contributed data, may
forth a searchab1e editable metadata and bask information repository for-Uns
1602, such as to
form a URLipedia. The repositoty may be linked to one or More social media
maps 1604.
102011 In an embodiment and referring to FIG. 17, clustering (e.g.,
relationship-based, manual,
attentive, or the like) of data from at least one social media community may
be used to generate
anaOtionable targeting list. Targeting lists combine network centrality 1704,
issue relevance 1708
and CFI for a cluster 1.710 into a ranked target list 1702 that may be used by
marketers or other
interested parties in order to reach certain nodes.: in some meaningfil order
for targeting for
strategic communication or other business purpose. The formula of combination
may be adjusted
to maximize ranking to suit client/user objectives. In an embodiment, network
centrality may be
a universal score related to how central a node is in the network. For
example, daytime talk show
hosts may have a network centrality of 100 in the general population, while
economists may be a
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
zero. In an embodiment, a Cluster Focus Index score may be calculated for each
cluster. For
example, daytime talk show hosts May be a zero:CFI:for economics, but
economists are .100. In
an embodiment, an issue relevance score may be calculated for each cluster.
For example, the
issue relevance related :to the budget deficit may be calculated based on a
publication 'frequency
-score (e.g., /1 of tweets). Other score techniqucs. may be used to calculate
an issue...relevance.
102021 in an embodiment,. -users may he able to purchase ads or message
placements on a target
from the targeting list 1712. From the targeting list, Users may be enabled.
to buy an ad placement
or message placement on the target site at the click of a button.. In an
embodiment, the effect, or
impact, of the ad/message placements may be tracked for the node and across a
social media map.
Thus, the system may enable users to identify targets according to a ranked
list based on network
centrality, CFI, and issue. relevance, and. then place and track ads/messages
on the targets from the
lists. In another embodiment targeting lists may be used:in-connection with
any ad network for
ad/message placement. Tracking: ads/messages may involve receiving feedback on
actions taken
with respect to the adsirnessages,_calculating imp,* *tries; and the like.
102031 in an embodiment, a historical data browser may-provide a mechanism for
Visualizing
Archived, historical social Media map data,-.400171,40 ft research or
historical purposes. For
-example, there may be value to acadernitof accumulating old social media maps
and showing the
delta between them, such as to explore_ how::the.market has evolved over some
period. of time
Historical social, media map data may also he awful. for financial industry
forensics and
intelligence analysis.
102041 In an embodiment, CFI metrics may be. displayed: on a .soCial Media
map. A CFI metric
for items inelasters indicates.hovv- much attention there is-to that. item for
that cluster._ A.n_-attention
score indicates the relative attention to an item as compared to other items
for a cluster for a range
of time or for a '.`point" in time. A higher attention score means the item is
more specific to the
.cluster. Attention scores are nen,-linear in the sense that anything below
two is not significant and
greater than two, it is exponentially significant,
102051 CH scores may be a metric for measuring search engine optimization
and/or advertising
effectiveness because it represents cluster specificity. CFI metrics would
have to be combined
with a more global metric to enable cornpanies. to shift from thinking at the
execution/implementation layer (egg., where dot advertise?) to the strategic
.layer (e.g., where are
we going with. this community?)._
102061 In an embodiment, atF1 Graph may include C}71. scores forsources and
nodes on the map.
In the upper right of the. map are clusters with high .fbetts on the
partitular cluster, high overall
level of attention, and many in-links. On the CFI graph, users can see various
hems at a glance.
46
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
For example, users may find the. key players related to a topic or the
landscape of players to
determine who has influence.
102071 In an .embodiment, a (.7:11 graph may include a Cluster Map Properties
Editor/User
Interface.. The interface enables users to label clusters, assign clusters to
a group, and perfOrm
group metrics.
10.2081 Maps may be generatedbased on semantic elements, bundles, white lists,
black USK :and
the like in an automated fashion in come embodiments but labeling the clusters
in an automated
way, such as when a map update is made, may be difficult. Draft labels may be
assigned when
the cluster is created or updated based one previous storehouse of knowledge.
.A confidence score
ti..4 to that labeling may be generated. To automate the labeling, members of
a cluster may be
compared with membership of clusters of past maps and if a high percentage are
the same then it
is assumed theclusters relate to the Same thing and are labeled similarly. In
another embodiment,
automated labeling is based on a structural equivalence. Labeling a node or an
object that has
well defined properties may be easier than labeling a. cluster, which is a
colleetion _of objects.
Structural equivalence involves examining the node's Winks. For example, if
people are friends
with the _same people, then they may have similar interests. In another
example, blogs that link to
the same sets of things are likely to be similar. In yet another example, :if
there are two people
who have superiorrelationships to twenty soldiers, chances are that die two
people are sergeants
or some otherform of commander. While this may work at the node level, it is
harder to do at the
cluster level. CR scores, which are already generated for clusters, may be
used in the generation
of labels. For example, for two clusters with numerous links from nodes in
these clusters to other
nodes, it is difficult to compare the clusters at face value. One might just
be larger, more popular,
or have more links. However. CFI scores enables a comparison between two items
or sets of items
that a Ouster may be disproportionately paying attention to. For example,
Cluster I is very
interested in horses and baseball,. while Cluster 2 is very interested, in
hones and basketball. Given
the CFI scores, vector cosine similarity can be used to determine
therelationship between the two
clusters. For each cluster, vectors can be built based on the CFI scores
calculated for each of the
clusters for the same items(e.g., Cluster 1C,F1:1.(1).õ-CFB(2) . etc.; Cluster
2=CF12(II.), CF12(2)
. etc.). The vectors may be plotted in a.30-vector space. The cosine of the
angle between the
two vectors May be one indication. Of therilatioriShip between the. two
clusters. If the eosine is
small, the confidence is high. As maps are updated with new Content,
thiSterSin the new map can
be compared to clusters of old maps. When there is a match, that is, a small
angle between two
-cluster vectors, the label from the cluster in the old map is assigned to the
cluster in the new map.
In embodiments, the .cosine of the angle may also act as a similarity score.
There are a number of
47
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
measures for vector distance, including correlation distance, cosine
similarity, Enclidian distance,
and the like.
102091 in embodiments, to limittheiiumber:ofcFl%.toinciudein vectorgeneration
the CFI's may
be filtered to include only a CEI.OftWo-or-more.ottaparticularclu,ster, This
effectively reduces
the dimensionality of the spat*
102101 In other embodiments,..itertis that are similar May be aggregated in
labeling: For example,
using outlink bundles rather than an individual CFI Score may enable grouping
items into target
clusters and examining the density of links to the target cluster.
101.1.1.1 In an embodiment, an advertising campaign planning tool can enable
running a campaign
on blogs, and.4*eklug 4i4ceeas in other In) ets (e.g.., TWITTER1":,
FACRIQOKTm.; segment-
specific onlitieforitinS),-
102121 In an ettibodiment, URL shorteners included in social media content may
be tracked. The
system may provide reporting outputs that track the success of a social media
campaign including
a URI, Shortener in different layers of the social media: system. The system
may not only be used
to plan the campaign, but may also. be used to report on the TWITTERT" bounce
from blog activity
Or the FACE:BOOM" bounce from blog activity, for example.
102131 in an embodiment, the system .may enable campaign planning (e.g.,
domestic,
international, multi-platform, multi-network,ete4 where language is not a
required first limitation.
For example, the system may enable campaign- planning in marketing, such as,
for consumer
goods, media and entertainment, movie marketing, video games, social games,
music,
international product launches, talent agencies, public diplomacy, public
health, political
campaigns, and the like. Campaigns may be tracked, such as with a
chronotopeanalysis, as will
be further described herein, to determine a pattern that exists in time and
space determined by
combining temporal and network features in the analysis of the
segments/clusters.
102141 In an embodiment, the system may marry internal reporting with other
reporting tools such
aisplash, resonance, clicks, transactions, and the like.
102151 In an embodiment, the system enables analysis and prediction, such as
in the financial
industry (e.gõ market predictions and trading positions), social media firms
whose value is built
around prediction, and the like.
02161 In embOdiments, third party data and clusters may be used with the
mapping techniques
described herein.
102171 In embodiments, models may be built on one or more clusters using tools
that can be
accessed across clusters.
102181 In some embodiments, a social media map and network segmentation may be
constructed
via clustering of data from a single user's social media community. Referring
now to FIG. 23, a
48
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
user flow for becoming a user and interacting with a map is. depicted.
Starting from logical block
2360; prodeSsintiki* proceeds:A a login screen at logical block 2302. where
users may log in,
such as via a social, media authoription. If the user is a new user,
the:::user is scut to a sign up
page at logical.b.100k 2304, where they may sign up or be given additional
content to entice a sign-
up. lithe user is already on a list as having requested access, processing
flow proceeds to logical
block 2308 tocheck a wait list status. If the user is a beta user, processing
flow proceeds to logical
block 2310:Where-it is -determined if the login is a first login. If so,
processing flow proceeds to
logical block 2312 where a tour may be taken. After the tour, processing flow
may proceed to
logical block 2318 where a map overview is presented, including a competitive
overview, a text
description, a cluster power, and the like. If the user is not a beta user,
procesSinelow may
proceed to logical block 23149 where the delta since last visit is presented,
including. new
followers, recent activity With niap indicators, and the like. Processing flow
may then proceed to
logical block 2318. From logical block 2318. processing flow may proceed back
to logical block
2314 if recent activity is requested again.
1021.91 Alternatively, if the user chooses a cluster or group at logical block
2318, processing flow
:may proceed to logical block 2320 to obtain a cluster overview, including
local competitive
.performance, influencers, conversation, images, videos, recent tweets, and
the like. If the user
chooses to delve into the entire interactive map, processing flow may proceed
to logical block
2322 for clustermap navigation. Processing flow may alternatively proceed to
logical block 2324
from logical block 2320 where the user may take action. In an alternative
embodiment, processing
flow may first proceed to logical block 2328 where the user may first view
full lists, and then
processing flow may proceed to logical block 23.24 where, only actions that
are relevant to the list
being reviewed are displayed at logical block 2324.. From logical block 2324,
the user may choose
to build a network, save one or more clusters as a list, move a message.
engage with content, or
the like. If choosing to 'build anetwork, processing flow may proceed to
logical block 2330,Where
the user is prompted to make a list of influencers. From there, user details
may be entered at
logical block 2332, and then actions such as engaging one of the users make
current logical block
-2.334-ora follow action may be taken at logical block 2338. From logical
block 2330, a follow
.list maybe generated at logical block 2340, or the current view maybe saved
as a Twittefrm list
or some other social media list at logical block 2342. Likewise, tribe.
",90Ø<31i00 *s*. List" action
10.:sete44-preirrnsing flow may proceed to sage tlie current view as a
TwitterrmligOt.Some other
social molialistailogical block 2342. If the move message action is selected,
a lista-followers
maybe made at: logical block 2344 and from there The current view may be saved
as a Tivitterm
list or some other social media list at logical block 2342, or a message may
be composed at logical
block 2348 which may include content and context and the message. If engage
with content is
49
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
chosen at logical block 2324, processing flow may proceed to logical block
2350, where a list, of
content, such as URILsõ key content and media, may 1* made. Users may choose
to. screen content
details at logical block 2352 after which processing flow may proceed
tological block 2300 where
a word tweet is generated, logical block 23* where a re-tweet is generated, or
logical block 2354
where tweets by influencers who tweeted thecontent are found and then
potentially re-tweeted at
logical block 2358:
102201 In order to scale the amount of information in the social media maps,
clustering techniques
may need to be modified. In general, some set of nodes pay attention to some
set of targets and
the nodes get Clustered basetkm the targets they pay attention to. There are
at least two extensions
of this general approach. In one embodiment, a very large number of nodes pay
attention to a
very large number of targets. Thus, for clustering, the number of operations
scales at least
polynomially (e.g., the cube of the number of nodes). For example, for 10,000
nodes the number
of operations is in the billions. To accommodate this scale, computing power
may need to be
augmented.
102211 In another embodiment, attentive gravity may be used to scale up the
size of the social
Media maps. Nodes pay attention to targets (input data), however an object may
be created where
nodes are -not discretely assigned to a cluster but are drawn to different
poles, such as ideological,
thematic; or topical poles. Depending on which nodes a target pays attention
to., it can be drawn
to one pole, another pole, or the..middle. Instead of discrete maps with a
plurality of clusters (e.g.,
40) in a plurality of colors (e.g., 40), an attentive.gravity map may have
poles where the nodes.are
distributed based on how close they are to each pole. A node may have a. Set
of scores which
represent a gravitational coefficient- for each of the .poies of gravity. The
gravitatio.nal coefficient
may be used with other visualizations in order to modify the size, color, or
opacity of the cluster
representation based on the attentive gravity toward a pole. In another
embodiment, the
gravitational coefficient may simply.he used as a metric on the cluster map
previously described
herein. The gravitational coefficient provides the degree to which a node
matches aseginentation
(e.g., a sports weight and a parenting weight for the same node, rather Than
just sorting the nodes
into different clusters/segmentations and throwing out the relationship .to
other clusters or
segmentations).
102221 (lusters themselves may not really be definitive. For eXaniple, a node
might not be in just
one cluster. Such characteristics may be reflected in mapping technologies.
102231 One technique may be a Discrimination Function. in an example,
1,000,000 nodes may
be elustered. An initial condition may be a seed attentive clustering for a
small number of nodes,
such as 10,000. To expand the clustering, the centroids of the clusters are
used to assign values
to the other clusters (the X, Y average of the dots). For example, it can be
determined if a new
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
node is closer to the centroid of one cluster or of another. As many nodes as
desired to be
incorporated into a map may be Clustered via this technique. In this example,
this technique
applies to nodes 10,001 through 1,000,000.
102241 Another technique may be to iteratively cluster the 1,000,000. nodes in
batches of 10,000.
Then, the CFI scores of those clusters may be used to cluster like clusters
with each other. The
clusters may be combined at a meta-elustet level. To make that Work well,. how
Similar some
clusters are may need to be tracked across large groups of sub-clusters to see
which ones OTC
idiosyncratic and should standalone versus ones that are somewhat consistent
and shOuld be
joined.
102251 In an embodiment, it may' be desired to reduce the scale of the map to
just those actors
connected at a mesoscale- while eliminating actors who. are = not really
active members of the
network and are just "star" Ibllowers. An Influence Network Discovery method
may be used to
reduce very large networks to their most influential core communities and
obtain a sub-graph of
.maximally connected sub-actors.. A variable Kcon. "nay be assigned. to each
member of the
'network, where KCOrf relates to a minimum connectedness, or the number of
other nodes in the
:network. to Which.. the individual is connected (e.g., a known measure of
connectedness in
networks); One way to reduce the network quickly is to restrict the network.
by le..coti value. For
-example, a network may be restricted to only those with a KCOIT of five and
tip, that is, only those
people connected to at. least five other people. Another way to reduce the
network may be done
iteratively. For example, a network of people surrounding the Democratic Party
may be reduced
iteratively. In a first step, inactive members and members with few followers
may be eliminated.
Then, certain network members, such as public figures or those who, have a lot
of followers may
be removed temporarily from the network and reserved in a "keep" set. Then;.
The remaining
network. may be examined and refined by &orr. In the example, members of the
.network with a
KauT of one are removed from the demerit. Removal Of these- people from the
network may
change the Kau,' values for the remaining members of the network. The. process
iterates, removing
those network members with the lowest Kcori values. The process can iterate
until a. specified
number of network members is obtained. At this point, any members in the keep
set may be added
back to the network. .A.S a second pass, a Ktvrt of the keep set members may
be done and I itnited
to the node threshold. Based on the follow patterns of the members retained in
the map, they may
be assiened to a cluster.
102261 In an embodiment,- a delta report may be provided to examine the
evolution of a Cluster
map, and capture the most sal lent points of change in the last interval. The
delta report may identify
which clusters have grown, which sites are being targeted more by clusters now
than before, which
topics webeing discussed more now than before, Which clusters are more active
than before, and
51
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
the like. The delta report may be provided on a periodic basis, such as
weekly, monthly, and the
like. Generating the delta report may involve reporting which CFI scores
changed the most and
which clusters are more active than before. Delta reports may be enabled by
organization into a
self-updating database with time snapshots. A delta report may be useful in
customizing a stream
()iconic/it', For example, a stream of new objects of interest fbr clusters in
the map can be provided
as .a delta report and feed to a user.
[0227j In an embodiment, a self-service tool may he designed to let users
access the system and
initiate generation of a social media map. In an embodiment, a user may log in
to the system or,
in embodiments, to a social network or other third party .website, in order
to. initiate the map
creation processõ4 hot maybe spawned that harvests data and mapsthe data to
clusters. The hot
may further provide cluster labels and CH .scores. The output may be a social
media map data
object with CFI scores, The self-service tool may enable user browsing of
chtsters and the map,
tagging nodes, grouping and labeling clusters, and the like. in an embodiment,
a machine learning
labeler may suggest cluster labels. The user-generated labels may be fed into
the machine learning
facility used to label clusters for the social media maps. The focus of the
self-service tool may be
On actions that strategically build a user's network and strategically message
to components of the
network, ens can be used to determine a similarity among maps so that an
existing social media
map that. is similar to the self-service map may be recommended for review.
102281 Social media maps may be used to enable users to strategically message
components of
their network. In an example, a social media map may be created for the
Twitterrm followers of
a live entertainment company. Certain clusters relate to dense communities
around particular stars
or particular genres. of music. For the live entertainment company, there are
relatively few
messages that they transmit that everyone in the map cares about however,
using social media
maps, clustering enables more discrete message targeting. if the company wants
to use Twitterml
to get the word out about a 'country artist, ler example, they can target the
country music cluster
only with their messaging. If the company wants to target only those nodes
within the country
music cluster that have the highest influence. Cfl scores may be used to limit
the messaging in
order to maximize the impact on the cluster. Such discrete targeting may be
particularly useful in
the case where direct messaging to followers may be limited.
02291 Social media maps may housed to enable users to strategically build
theirnetWork. FOr -
example, in the live entertainment company, the country music. cluster may be
growing in. size.
The social media map may be used to identify niche. influential nodes for the
country music cluster,
such as by using segment CFI data to maximize connections- with targeted
segments/key
influencers. Then, the user can start following those influential nodes in
hopes that they will
follow back. Such a process may help build the network in a desired strategic
direction. Users
52
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
may be able to see how they are doing against competitors for any given
segment by examining
the proportion of influeneerS (high CFII target), who may -or may not be in
the map; following them
-versus others.
102301 In one embodiment, social media maps may be organized and navigated as
a imp of maps,
where each map appears as a node on a larger map. The strength of the
connection between maps
is the maximum of ratios of how. many nodes are in one map versusunother Map.
In navigating
and searching the maps for a particular target, an indication may be given
when a cluster in one
map is very -similar to another cluster in another map that may or may not be
accessible by the
user,. for example, if one map relates to diabetes and another relates to
obesity, a common Cluster
May. be :groupS iletiVety modifying lifestyles to avoid both pathologies In
embodiments, the
system may provide an interfitee from the search, screen with which the user
may purchase the
map they do not currently have access to.
102311 In an embodiment, user segmentation may be used to find segments for
targeting as
customers. Maps may be automatically generated for the target customer and
conversion rates to
paying customers may be tracked.
102321 Described herein is a system fir examining social media phenomena, such
as hashtags,
and how they spread in a network. Patterns of' spreading may include salience;
commitment, or a
combination thereof termed resonant salience, where them is a burst of
activity 'followed by a
sustained commitment,. or resonance, pattern. By combining temporal and
network features in the
analysis of the segments/Clusters, chronotopes (i.e., patterns that exist in
time -and space) emerge.
(02331 In an embodiment, a timeline view may be used to examine messages
across clusters, The
timeline may include the chronotope- as the drill down. For example, a primary
timeline maybe
organized in rows by grouping of clusters (e.g., similar clusters are assigned
together into a group).
'There may be several bands for groups (e.g., things for which there is-a CFI
score). The timeline
may be examined for objects of interest that _ have very high C.F1 .scores at
some point, One
example may be hash tags in a Twitter network. A dot May be placed at the
point in time when
the activity (attention) peaked (had the most citations, re-tweets, etc.) for
that object of .interest.
A dot may be placed in the macro timeline for the group (showing the peak
points of all objects
of interest) where. the peaks were for each group (a group corresponds to a
band. below the =it)
When the dot that 'corresponds to the peak of attention to an object of'
interest 'for a
groupfcluster is clicked, the chronotope is revealed. The chronotope for that
object Of interest
may appear in a window below the timeline. The timeline view may include time
on the X axis
and groups/clusters on the Y axis. Peak interest points forobjects may appear
as dots at points in
time corresponding to the groups that have interest:, Clicking on that object
reveals the chronotope
for that object for all of those groups.
53
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
192341 interacting with data in the chronotope -view may reveal what the
object Of interest is. In
Some embodiments; a group of items may be selected at a time period for .a
certain cluster/group
and a word cloud or semantic analysis of prayer nouns that appear in those
items may be
assembled.
102351 Social media sites enable users to engage in the spread of contagious
phenomena:
'everything from information and rumors to social movements and virally
marketed products.. For
example, Tvvitterm4 has been observed to function as a platform for 'political
discourse, allowing
political movements to spread their message and engageSuppOrters, and also as
a platform for -
information diffusion,.allowing everyone. from mass media..to eitizcits-Wmach
a wide auclienoe
with a critical piece of flews. Different contagious phenomena may display
distinct .0)0*(6On
dynamics, and. in particular,. news may spread differently through a
populatiOn than: other
phenomena. Described herein is a system for classifying contagious phenomena
batied.onlhe
properties of their propagation dynamics, by combining temporal and. network
it.atures......klethods
and systems described herein are designed 'to explore the propagation of
contagioustiadnags. in
two dimensions: their dynamics, that is, the properties of the time series of
the contagious
phenomena, and their dispersion, that is, the distribution of the contagious
phenomena across
'communities within a population of interest. Further described - is A method.
for simultaneously
visualizing both the, dynamics and dispersion of particular -contagious
phenomena. Using this
method, particular contagious phenomenon: chronotopes, or persistent patterns
across time and
network structure, may help emerge a taxonomy for contagious phenomena in
general.
(0236i Given some contagious phenomenon p, p may be considered to have spread
to user u the
first time that u engages with p. For simplicity, engagement is measured as.
mentioning the
phenomenon. For news, mentioning is likely a sufficient form of engagement,
while for a political
movement, stronger evidence of engagement may be preferable (contributing
money, attending a
rally, etc.). 'IlOwever,.. in-sOcial media sites, hightrlevels of mentioning
often correlate withhigher
levels of engagement (e.g., users_ tweet. about a political rally), while_
false indicators of
engagement are rare: if a user wishes to mention a political movement, to
disagree with it, she will
.often not use a tag or specific, name, referring to that Movement, but. use a
variant of it. (e.g., a
TWitterm user who wants Vladimir Putin out of power may use thetag #Patinout
instead Of OPutin
when tweeting, about the prime Minister and future Russian. president).
Therefore, the number of
first mentions of p by users in some social Mega site is used as a proxy for
the number of users
that p has spread to.
102371 in an embodiment. measures for characterizing contagious phenomena
propagating on
networks may include peakedness, commitment (such as by subsequent uses and
time range), and
dispersion (including-normalized concentration and cohesion).
54
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
102381 The peakedness of a contagious phenomenon is a scale-invariant measure-
of how
concentrated that phenomenon is in time. A peak may be defined as a day-long
period where total
.first mentions by day lies twastandard deviations above the median. first
mentions. The specific
duration of the peak window and the required deviation can be varied to
maximize usefulness for
particular .kinds of phenomena and for particular social media networks.
Median may be used
instead of mean because, due to the skewed distribution, of first nientionS by
day for most
contagious phenomena, the Mean is over-inflated. Contagious phenomena with
short lifespans
tend to have a sharp4iM4 When a large number of people: mention the
phenomenon, but the
number of mentions is verysmall on either side of the peak. in centrast, long-
li fespan contagious
phenomena tend to groW::.'SlOWly, with a less pronounced peak of mentions: The
peakedness of a
contagious phenomenonit the *action of all engagements with that phenomenon_
that occur on
the day with the most; eneagements_with that phenomenon. A high peakedness
means that most
of the network's eneagement with the phenomenon (e.g., for .a social network,
people in the
.network mentioning it). occurs within a short span of time, typically, hours
to days. in contrast,
low peakedness means that the network's engagement with the phenomenon is
spread over a long
:period Of time, typically, weeks to Months. Phenomena with high peakedness,
such .as news
stories,. may propagate rapidly through the network and them dissipate justas
rapidly in the course
of the daily news cycle. Phenomena with low peakedness may include popular web-
sites_ and
videos, which may maintain a slow but steady rate of engagement¨individnals in
the network are
constantly discovering these phenomena, even as others get tired of them and
stop engaging.
102391 Commitment is the measure of the average scope of engagement: with a
particular
contagious phenomenon by nodes in the network, or the staying power of a
phenomena. Using
the example of people engaging with online content in a social network, the
commitment with a
particular piece of online content can be the average scope of mentions of
that content by pieces
of the network. This Measure would, for-example. differentiate -between
apolitical movement that
is just a fad, anti:brie that accumulates a number of diehard supporters who
keep the movement
alive. Scope may be measured in at least two ways, which leads to the
following two sub-
measures: Commitment by Subsequent Uses-and Commitment by Time Range in social
media
sites, the cost in terms of time and effort to mention something fen-Me-so:0nd
or thied or tenth
time is relatively small; thereforei for a -SceOrid dimension, tWo quantities
may be defined: first.
the average number of StibSequent mentions (all Mentions excluding the first_
mention of the
phenomenon by a user)ot a contagious phenomenon among the adopting usem-and
second, the
average time difference (in days) between first and last mention of the
phenomenon .among:the
adopting users. While the first measure, "Commitment by Subsequent Uses," is
relatively easy to
inflate by mentioning the phenomenon multiple times in a short period, 'the
second measure,
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
"Commitment by Time. Range", indicates long4ertn commitment to mentioning the
phenomenon
by a set of users.
102401 Commitment by Subsequent "Uses is the average number of subsequent
engagements with
a phenomenon after a node's first engagement. For instance, if each person. in
a social network
played an online Rime at most once, Commitment by Subsequent Uses tbr that
story would be
zero. In contrast, ifjust one percent of the people in a social network played
an online game thirty
times each, Commitment by Subsequent 'Uses for that game would be twenty-nine.
Phenomena
with high Commitment by Subsequent Uses may include online games, which
encourage repeat
engagements. Other phenomena with high Commitment. by Subsequent Uses may
include fiSt10-
tutfed content, where a third party May encourage repeated interest in the
cOrdent-..by paying or
otherwise endorsing people who engage with it.
102411 Commitment by Time Range is the average time period between the first
and last
engagement with a phenomenon by nodes in the network, measured over some large
time window
(e.g., a year). For example, if each person in a social network read -
.articles on a blog ten times
over the course of one day and never visited it again, Commitmentby Time Range
for that Wog
would be one day. However, if just one percent of the people in a social
network read articles on
a Woe once every week for ten weeks and then abandoned it, Commitment by Time
Ranee tbr
that bloe would be ten weeks. Phenomena with high Commitment by Time Range
include blogs
with loyal followers who keep coming back. for more content. Phenomena with
low commitment
by Time Range include news Stories that, on average,. a person reads .only
once and never sees
again.
102421 In addition to measuring the dynamics of contagious phenomena (the
properties of the time
series of engagements with a phenomenon), the dispersion of contagious
phenomena (the
properties of distribution of a contagious phenomenon throughout a population)
may be measured.
Dispersion is a. measure of the diStribution of engagements with a contagious
phenomenon over
the network through which it propagates. Phenomena that are highly dispersed
are broadly
popular but may have less focused engagement from a particular group;
phenomena that are not
dispersed are not broadly popular, but may have focused engagement with a
particular group.
There are many ways of measuring the distribution of engagements with a
phenomenon over a
network, including the following two sub-measures: Normalized Concentration
and Cohesion.
102431 The Normalized Concentration of a contagious phenomenon presupposes a.
partition of the
underlying network into discrete clusters, which usually represent
communities. Given such a
partition, the Normalized Concentration of a. contagious phenomenon is the
fraction of all
engagements that come from the cluster that engages. most with the phenomenon,
or the Majority
Cluster, For instance, if a social network were divided into two clusters, one
of which engaged
56
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
with a particular news story nine times, and the other, only once, the
Normalized Concentration
for that phenomenon would be 0.9. However, if both clusters had engaged with
the story five
times, the Normalized Concentration -for that phenomenon would be 0.5,
Phenomena with high
Normalized Concentration tend. to be the cause cekbre of a particular
community, e.g., political
and social movements that have not gained wide traction. Phenomena with low
Normalized
Concentration may include headline news stones that touch many communities at
once.
Depending on the size of individual communities, Concentration may or may not
correlate
inversely with popularity.
(02441 In addition.-to Normalized. Concentration, some aspect of the
connections between the
engaged users may be measured. For example, it is possible that A contagious
phenomenon is
widely spread across a number of communities, but diffuses only through strong
ties so that the
engaged users form a clique. Conversely, it is possible that a contagious
phenomenon is confined
to a single community. but spreads through weak ties and the engaged. users
are sparsely
interconnected. Therefore, a measure of Cohesion may be defined as the network
density over
the subgraph on all users engaged in a particular contagious phenomenon.
Contagious phenomena
that spread over strongly connected sets of users will have a Cohesion close
to one, whereas
phenomena that spread over weakly -connected sets of users will have a
Cohesion close to zero.
The Cohesion of a contagious phenomenon is the network density of the sub-
graph of all nodes
engaging with. the phenomenon. The network density of a graph is the total
number of actual.
connections between nodes in the graph divided by the total possible number of
connections
(usually n*(n-1.)/2 for undirected graphs, where n is the number of nodes in
the graph). For
example, if only three .people read a particular blog., but all those people
knew each other, the
Cohesion of that blog would be 1Ø In contrast, if ten people read a
particular blog, hut every one
of those ten people knew exactly two Odle others (the people were connected in
a circle graph),
the Cohesion .of that blog would be I0/(10*9/2)=10/45-0.22, Phenomena With
high Cohesion
may include stories and MIMS that propagate in an "echo chamber" of peoplewho
already know
each other and engage with similar kinds of online content. Phenomena with low
Cohesion
include news and rumors that move between acquaintances, such that, for
example, after multiple
propagations, the person who hears the rumor and the person who started it may
be total strangers.
(02451 In embodiments, phenomena with high Peakedness tend to have low
Commitmentõ .making
those two measures a natural pair for comparing different online phenomena.
For example, FIG.
18 depicts Commitment by Time Ranee. on the 'Y. axis and Peakedness on the X
axis for two
different sets of data depicted by different icons. In this example, the two.
datasets are: 1.) 112
Bundled .hashtatts relating to specific topics shown in red or as icon #1; and
2.) a baseline dataset
of the top 500 hashtags for all users shown in black or as icon #2. The
bundled hashtaes.display
57
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
a generally lower level of Commitment by Time Range than the top 500 hashtags
at the same level
of Peakedness. Some of the top 500 hashtags have extreme levels of Commitment,
up to 150
days. Hashtags with the highest levels of Commitment areof several sorts.
which notably include
regional/location tags, tags for particular sports, religion tags (e.g.,
"Catholic," "Jewish"), tags for
particular news outlets, and general tags related to investing and financial
markets. Intuitively, all
of these are topics.that might engage a stable set of users Over a long time..
102461 Referring to FIG. 19, and in an example, dealing primarily with topics
related to Russia,
'peakedness is plotted for the bundled hashtags against both levels
ofrommitment: subsequent
uses XPICI. 19a) and time range (FIG. .19b).. in Fla 19a, there are several
diStinct. regions of the
diStribUtion. On the bottom right, hashtags. with high Peakedness-and low
Commitment by
StthS4tent Uses are all directly related to salient news events, which in this
case are the airport
and.--rnerek bombings in Russia (#Domodedovo, #explosion, #inetro2),
#Moscow29). On the
bottom. latt; hashtags with low Peakedness and low Commitment by Subsequent
Uses are
Betterally nOt very popular. Some of them are very generic (#moscov4 #rnetro),
and some just
never had a peak nor became adopted by a committed user base. Some elf these
are tags that are
similar to popular tags, but reflect less-used variations. On the top left,
bashtags with low
Peakedness and high Commitment by Subsequent Uses are all regional hashtags
(with the
exception of the Nashi hashtag that refers to a pro-government political..
youth movement in
R.ussia). These regional hashtags were tangentially related to the forest tire
events, but their main
use is likely in talking about local atihirs, hence the high commitment eta
few users. Finally, on
the top right, there are a number of hashtags with both high Peakedness and
high commitment by
Subsequent Uses. These tend to be pro-government political hashtags (#i Ru and
#GoRtt are both
related to Medvedev's policy of modernization while #ruspioner and #seliger
are both related to
the Seliger youth camp). This observation suggests that pro-government
political hashtags have
some event (such as the Scliger camp) that is linked to a sudden burst of
popularity, but subsequent
to that event,- Users Continue to include the hashtag in their tweets. This
suggests that pro-
government political hashtags may have "staying power" in the Russian Twitter
community.
Alternatively, or in combination with this, a. committed set of users may use
the pro-government
hashtag both before and after the event, perhaps in an organizational or
mobilizing capacity.
102471 In contrast, and referring to FIG. !-19b, some of the sante clustering
seen in FIG. 19a is
depicted, where news is on the bottottt..right, regional hashtags are on the
top left, but. the top right
group dominated by pro-government hashtags has moved down, indicating that
these hashtags do
not have stayingpower overlong periods of time; they may be mentioned multiple
times, but in a
relatively short time range around the peak (days or weeks, not months). In
contrast, the hashtags
on the top right in FIG. 19b are the regional hashtag #Moseow and the
political hashtag #Putinout
58
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
(referring to the anti-Putin movement). Alls-.important to note that -
#Putinout in particular has
relatively long temporal staying power (anavetage of 50 days between first and
last mention by a
user in the clataset) but relatively short staying power by mentions (an-
average of less than six
subsequent mentions).
102481 Referring to FIG. 20 and FIG. 21, measures of dispersion of hashtags
are analyzed across
a core set- f Twitterm4 users. In FIG. 20, 'the distribution across nine
topics .of Normalized
Concentration are plotted by hashtag within each topic. Comparing across all
nine topics enables
distinctive patterns to emerge; the. minimum Concentration among pro-
government hashtags in
the Seliger and modernizatiOn:- topics is between 0.3 and 0,4. In contrast,
the maximum
Concentration among opposition. .hashtags in the Kashin and Russian Drivers'
Movement topicS,
is between 0.4 and 0.5. Pro-government hashtags are on the whole more
concentrated within one
cluster than opposition hashtags. Hashtags related to news events, such as the
Moscow Metro
Bombing and the Domodedovo attack, tend to be diffuse, which is in line
with.the intuition that
.major pews events tend to engage the population as a.whole rather than
specific communities.
102491 inFiG. 21, the distribution across nine topics of Cohesion are plotted
by hashtag within
each topic. For ease of visualizing, the distribution plots'are cut off at 0.2
and all hashtags with
Cohesion >0.2: arc assigned a value of 0.2. Again, there is a contrast between
opposition bashtags,
which. have extremely small 'Cohesion of 0.03. and below, and some pro-
government hashtags
(especially those in the Seliger and modernization topics), that have the much
higher Cohesion of
0.10-0.30. Curiously, a few news-related hashtags have very high Cohesion,
which suggests that
some news-Mated hashtags may spread through strong ties.
j0.2501.FICIS. 18 through 21 provide a high-level analysis of hashtag.
diffusion among the
Russian-speaking Twitterm community, both from the temporal and the spatial
(network)
perspective. However, this analysis necessarily leaves out the idiosyncrasies
of individual
hashtags. Referring now to FIG. 226, FIG. 221,, and .FIG. 22c, chronotopes of
the ihnetro29 (a),
*samara (b), and ARti (e) hashtags are depicted. In typical. chronotope
images, color indicates
cluster group, and color brightness indicates volume of engagements. Detailed.
analysis of
individual contagious phenomena enables crossing the dimensions of dynamics
(loosely, temporal
properties) and dispersion (loosely, spatial properties) of the latter.
Therefore, spatiotemporal
analyses of contagious phenomena, such. as hashtags, may be constructed, and
patterns in their
diffusion across time and space may be discovered. Such patterns may be
called, the chronotopes
of the hashtags. A chronotope is simply a pattern that persists across a
spatiotemporal 'context,
originally used in literary theory to describe genres or tropes.
102511 in order to discover hashtag chronotopes, the diffusion of individual
'hashtags is visualized
both across different communities and across time. First, a particular hashtag
is selected and the
59
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
set of engagements of Twitterlm users with this hashtag is binned by day.
Next, for each day, the
volume of engagements for that day is broken down by cluster group. Finally; a
grid where
columns correspond to cluster groups and rows correspond to days is created.
Each row-column
cell of the grid is filled with a color corresponding to the cluster group. A
cue-as to the volume of
engagements corresponding to a particular cell is given via the brightness of
the color: the brighter
the cell, the More engagements with a hashtag on that: day. from that cluster
group. Black cells
correspond to days when a particular cluster group has no engagements with the
hashtag.
10252) FIG. 22 shows three such visualizations: the #metro29 hashtag related
to the Moscow
Metro bombings on Mar, 29, 2010; the #samara hashtag related to the Russian
city of Samara;
and the #iRti hashtag, related to President. Dmitri Medvedev's policy Of
modernizing Russia.
These three visualizations display three distinctive patterns across space and
time; #metro29, in
FIG. 22a has a "salience" chronotope, with engagements across the spectrum of
cluster groups
during the week around March 29. :In contrast, #samara in FIG. 22b has a
"resontmce" chronotope,
with consistent engagements. from the local cluster group, presumably
residents of Samara talking
about their city. Finally,...#iRg in FIG. 22a has a "resonant salience"
chronotope, with an initial
cross-group burst of activity in late November 2010 (around the time of
Medvedev's
announcement of his new policies) .followed by consistent -engagements from
the Pro-
Government cluster group over the nextmonth. Note that Fla 22 does not
contrast with FIG. 19,
which suggests that pro-government hash tags have low staying power, but
instead presents a more
subtle picture; the cluster group of pro-government users remains active-in
the 4iR u hashtag over
the course of a month, but, as FIG. 19h indicates, individuals within that
cluster rarely carry on
with adoptions for more than 5 days. There may belt high turnover of users of
the #iRti hashtag,
with new enthusiasts coming in even as the original, adopters lose interest in
the topic.
192531 In embodiments, phenomena with the Salience Chronotope tend to have
high Peakedness
and link Commitment, While _phenomena with the ResOnatice Chronotope tend to
have low
Peakedness and high Commitment by Time Range. Phenomena with the Resonant
Salience
Chronotope tend to have both high Peakedness and high Commitment by Time
Range.
102$41 In :an embodiment, a flexible. algorithm may he used for optimizing a
targeted network
influence campaign. For example, a user may have a high CFI' score, but they
may not Message
their social networks frimpently, thus targeting theschidiViduala may not
optimize the vativitigm
The algorithm may output an M Score, which may be calculated from a CFI score
plualOtneOttler
network or behavioral metric. In embodiments, wherever it is described to use
the CPI score, the
M score may instead he used to maximize campaign effectiveness: In
embodiments, the. M score
may bean interpolation of the numberof followers of the target_ item
(influence) arid:the:CFI score
of the target item (specificity). This mathematical calculation may result in
.a...00.r.ttuilized..sOre
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
-on a scale, such as a scale from 1 to 10 where. I is low impact and 10 is
high impact. Thus, the M
score is a general measure of influence and specificity.
102551 One way to calculate theM score is to combine Gland count, where count
is the overall
number of members on the _map that: have engaged with that target, in a
formulaic way. The
formula is M score¨count (alpha) (71=(1-alpha) [normalized 1 to 10).
102561 In embodiments, the M-score May be user-tunable, so that there is a
choice to prioritize
"segment specificity" -vs. "global footprint," and/or "network position" vs.
"behavioral profile"
(e.g., someone Who rehveets frequently) when selecting behavioral and/or
network metrics to
calculate the M score. In an embodiment,. for example, a slider 2902 may be
provided to users so
that can Select a target. that is more niche or more global. The M score
enables optimizing a
campaign on network position or on behavior. If the slider is dragged towards
"niche," alpha
approaches zero and the M score isnear equivalent to just the CFI score a: The
target item (high
specificity). If the slider is dragged towards "broad," alpha approaches I so
that the .M score is
.near equivalent to just the number of .followm, of the target: Item (high
influence). Setting.the
slider somewhere. in between -"niche" and "broad" allows users to tune the set
of
indiVidttalsienfitiesthat they want to target.
.102571 In an embodiment, direct ad placement may be enabled by CFI scoresiNi
scores. Using
C.F1 scores. and/or M scores, a list of targetstwebsites. may be created and
ads may be placed
directly on the target/website via integration with various products, such as
T.witterrm sponsored
tweets, Facebook"A ad. exchange, Googlerm AdSenselAdwordsõ third party online
ad networks,
and the like.
102581 Referring- now:to:FIG. 24, a recent activity page of a social media map
platform provides
recent activity, such as new Ibllowers, new influencers following the user, an
indication of any re-
tweets including the number-of people who have retweeted an item, changes to
the user's cluster
groups with links to respective group overviewsereeris, a list of new
influencers including their
cluster group and their number of followers,. the current conversation leaders
including their
cluster group and their number of followers, a view of all media being Shared
in -the network
including the latest influential medittand the segments M which the media is.
influential., links to
an overview page, links to a lists page, links to a help and support page, and
the like. The user
may continue to their map from this screen, .:Graphics, such as a bar graph,
may be included in
the changes to the user cluster gt'Ottbox .tdindicate the number of users in
each cluster group.
Graphics, such as a bubble chart, may also be included in the media box to
indicate the-size-of the
segments in which the displayed latest media is influential.
102591 Referring now to FIci..25, another example of a recent activity page of
-a social media map
platform is shown. In thiS.!example, new followers are shown; 'minded in the
number of followers
61
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
are new influencers and group changes, including a percent change for each
cluster group,
information on new influencers, such as their name, handle, number of tweets,
number of
followers, number of people they are following, and a button to message them
or follow them.
Also on this page are trending termsit.IRLs,inchtding the number of mentions
of a hasittag that is
related .to the user, trending media and imagery, and latest influen.cer
tweets. Icons may be
provided to reply, retweet, favorite a tweet, share or embed a tweet, and the
like.
[0260i Referring now to FIG. 26, an overview page is shown. The overview page
includes a table
of Cluster groups, the number of members in the group, the power of the
cluster, and the tweet
activity. A power score is an indication-of which segment is worth engaging
with and may be an
indication of which segments are most dense and represent the greatest signal
of interest. In one
embodiment, power may be calculated based on network density: the number of
connections
divided by the number of possible connections. In another embodiment, power is
calculated based
on coordinates, such as the average distance .from the center of a cluster
map. In another
embodiment, power may be calculated as the average distance from the centroid
of the cluster that
emerges in the clustering computation. In embodiments, power is like the
segment/cluster version
Of the M score.
192611 Continuing with the page on FIG. 26, an individual cluster may be
selected and a
representation of that cluster in a map maybe. highlighted. For example, the
UK. design cluster
has been highlighted and a dialog, box appears showing more int-imitation
about the individual
group, including number of members and graphics depicting the power and tweet
activity
associated with the group. When the user dicks the "Read more" link, a box may
appear with
more information. The map and group information items may remain visible when
the page
scrolls such that they are in a fixed position. Selecting clearer on the page
overview causes the
selected row to be cleated and makes all. map nodes visible. An alarm icon on
the overview page
allows the user to review all recent activity including, number of tweets from
various members Of
the network. Selecting "View full-screen map" will send the user to a screen
such as that shown
in FIG. 27. Referring now to FIG. 27, a full-screen map is displayed. In this
map, the international
cluster has been selected and the South America sub-cluster was selected. The
colored nodes in
the map may indicate one or both of the selected clusters and:sub-clusters:
The influencers in a
particular sub-cluster may be Viewed and when an influencer is .selected,-the
URIA associated with
that influencer there may be shown. A node overview may appear including the
influencer name,
their handle, their location, their EJR.L, when they joined 'the social
network, their number of
tweets, their number of followers, the number of people they are Mowing, the
groups they are
linking in, the number of in-links in each group, as well as any other
relevant information.
62
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
102621 Referring now to fla 28, an embodiment of an overview page is shown. In
this.. view, a
segment ;or cluster has been selected and data regarding that segment is
displayed, such as key
influencers, current conversation leaders (mentions), an interactive map, key
photos and videos or
other media, key tweets/retweets, key websites, key content, latest
conversation terms, and the
like. Effectively, this page shows an enhanced version of cluster-focused data
and makes it more
accessible. The power score for the segment is displayed as well as an icon
from which the user
may take certain actions such as build their network, find content, find
media, find tweets, message
followers, launch a. MitterTht campaign, launch a FacebookTm campaign, launch
a mobile
campaign, launch a social. media -campaign, launch an Ad Words campaign,
launch an
advertisement campaign, and the like, 'The oVerview page may be a user
interface. Notifications
of certain data and data presentation may be made in the user interface, for
example, which may
be implemented by software and embodied in a tangible medium, such as a mobile
device,
smartphone, tablet computer, or the like. The user interface may be a
touchscreen embodiment,
such that to utilize the user interface., a user is required to touch the
screen of the device displaying
the user interface. The user interface may be accessible on different
computing devices and
capable of dynamically accessing user specific data stored on a network server
and/or local device.
102631 Referring now to FIG. 29, tho"influencers" tab has been invoked.
Various ways to filter
the influencers are provided such as by follower status (all followers; &Rows
the user, does not
follow the used or by &flowing status (show all, the user follows; the user
does not follow).
Another way to fitter influencers may be by MI seem, follower- count,
Mentions, name, screen
name, and the like. One way to filter by M score is by
uscOrasijd60902.:to,obtain more niche
or broader individuals/entities as. described elsewhere herein. Another way to
filter-
individuals/entities may be by their exposure to particular content. By
utilizing this filter, the user
may target individuals/entities who have not already been exposed to the
content. Users may take
action from this page such as to follow selected indiOdualtientities, save
individuals/entities to a
Twitter"' list, create:a new list, add a selection to a I iSk-:Send a.direet:
message, send a sponsored
. . . .
tweet, and the like. When saving individuals/entities to aTwitterTm:list, a
dialog box may appear
with list Ohoices for the user, such as a list for my influencers:following
me, a list for my
influencers and not following me, a branding group, and the like. In this
example, one action
being taken is to follovv seven new users. By following individuals/entities
and engaging in
behaviors that might cause them to be awareof the user, the users network may
potentially expand
to include the newly followed individuals/entities. Another action that is
taken it to compose a
messatte. The compose message screen. may include suggested content: such as
most used hashtags
or other media based on a CFI, popular terms, key content such as high M score
media, and the
like. Influencer information may be leveraged in determining whom to message.
The suggested
63
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
content may be filtered by the exposure of target individuals/entities to the
content. Data related
to the content: such as its peakedness, first appearance, and the like may be
exp-osed to the user so
that the user- ctixt decide. whether it makes sense to share the content with
other individuals/entities.
Referring to. FIG.. 30, users may be able to drill down .to the individual
influencer level to see in
what other segments/clusters the individual is influential, their latest
tweets,. M score, number of
tweets, number of folloWers, number:following, footprint, followin-gifollower
status with respect
to the user, demographic information,. URL, and the like. Icons may be
available to follow, act
(he., add the person to a list, retweettheir latest tweet, send a direct
message, etc.), view asocial
media profile, and thelike,.
102641 Referring nevii.:tb..IFIG-11.1,õ a:tab:for conversation leaders -1'k
diSplayed. Various 14*$ to
filter the conversation leaders are proVided such as by follower status (all
followers, follows the
user, does not follow the user) or by following status (show all, the user
follows, the user does not
follow). Another way to filter conversation, leaders.is by peak date such as
all, today, past week,
:past meetle.cestem date range,. and the like. Another way to filter
conversation leaders may be
:by ;.M-:;aeore, follower count, mentions, peak, -peakedness, name,, screen
name, and the like.
Another way to filter conversation leaders may be by their exposure' to
particular Content. By
-utilizing this filter, the user may target individuals/entities who have not
already been exposed to
the content. Users: may take action from this page such as to follow selected
individuals entities,
save individuals/entities to e Twitter m list, create a new Iist, add a
selection toalist, send a direct
message, send a sponsored tweet, and the like.
102651 Refereing now to FIG. 32, a tweets tab is displayed. The tweets May be:
filtered by peak
date such as all, today, past week, past month, custom date range, and the
like. The tweets may
be filtered by M score, re-tweets, original postdate, peak, peakedness, name
of poster, screen name
of poster, and the like. One way to filter by M score is by use ota slider to
obtain an audience
that is more niche or breeder, as described elsewhere 'herein. Data regarding
each displayed may
include an M score the number of influential re-tweets, the number of retweet,
the posted date, the
peak date, a graphic of the peak pattern, icons with which to. -take action
such as
reply/retweet/favorite, name, screen name, and the like. Selecting one of the
tweets may cause a
drill down box to -appear with additional information about the
individual/entity who made The
tWeet,.. such as M -score,. number of 'tweets,. number of thllowees.,.,Mimber
following, footprint
number of friends, follower/following status, demographic data, VRL, which
segments' the
individual/entity is retweetingin, who have they been retweeted by, icons to
social media profiles,
icons with which to take actions such. as reply/re-tweet/favorite/add to list
and the like.
(02661 Referring now to FIG. 3:3,a websites tab is displayed. The websites can
be sorted by
mentions, M.score, subpages mentioned, hostname, and the like. One way to
filter the websites
64
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
by M score is by use of a slider to obtain an audience that is more niche or
broader, as described
elsewhere herein. Users may take action from this pagosuch as to buy an ad:,
create a new Iist,
add a selection to a list, and the like. Selecting-a webs ite reveals a drill
down box for the website.
Information about the website ln the drill down box may include M score,
distinct mentions,
mentions, subpages mentionett, excerpt, peak date, a graphic of the peak
pattern, segments/clusters
the website IS mentioned inõ. who mentioned the website, latest tweets
Mentioning this URL,
button to take action, and the like.
.10267i Referring now-to:MG. 34, a tab for key content may be displayed,.
Information about the
-Iseycontont included in-this view includes the name of the website, name
ofttnaiticle, URL, peak
date a peak. pattern. M score, citations, distinct citations, and thelike. =
The key content may be
sorted by Macore, citations, peak, peakedness, host name, content title and.
the like. One way to
filter by---.M Score is by use of a slider to obtain an audience that is more
niche or broader, as
described elsewhere herein. The key content may be filtered. by peak date
such.as all, today, past
week, paatiriOndi, custom date range, and the like. Users may take action from
this page such as
to composoaniessage, compose a tweet, view a drill down box for the key
content, and the like.
In the ôó pose message or compose Tweet view, users may be able to select one
or more
individuals/entities or and influencers/conversation leaders to message with
suggested content
(most used hasinags, popular terms, key content, etc.), In one embodiment, the
individuals/entities
may be part of a list such that either certain members of the list or the
entire list may be easily
included as recipients of the message. Selecting a key content reveals a drill
down box for the
content. Information about the Content in the drill down box may include name
of website, title
of article, M score, distinctinentions, mentions, subpages mentioned, excerpt,
peak date, a graphic
of the peak pattern, segments/clusters the content is mentioned in, who
mentioned the content,
latest tweets mentioning this URL, most used hashlags, a button to take action
(tweet this, use in
direct message, add list, eto.),:and. thelike.
1026/11 Referring now to FIG. 35, a media tab is displayed. Media may be
filtered by images,
videos, audio. Of's, and the like. The media may be filtered by peak date such
as all, today, past
week, past month, custom date range, and the like. The media may be sorted by
M score, citations,
peak, peakednesS, host name, content tide and the like. Information about the:
media in this view
may include title, duration, media type, M score, Mentions, distinct mentions,
peak date, peak
pattern, and the like. By selecting one of the media items, it drill down box.
may appear.
Information in the drill down box may include title of media; UlitM score,
mentions, distinct
mentions, peak date,: peak pattern, media type, duration,
whataegments/elusters the media is
mentioned in, most used hashtags, who has mentioned the media, latest tweets
mentioning this
media, an icon to take action with, and the like.
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
102691 Referring tO.FIQ:.14,.a tab for terms is displayed. The terms may be
filtered by hash tags,
one word, 2 words, 3-weeds, artdibe like. The terms may befiltered by peak
date such as all,
today, past week, past month, custom date range, and the like.- The terms may
be .sorted by M
score, citations, peak, peakedness,.hostriame,content title and the like.
Information about terms
in the list may include the term, peak date, peak pattern, M score,. mentions,
distinct mentions, and
the like. Selecting a term may reveal a drill down box. where additional
information about the
term may be displayed including which segments/Wasters the term has been
mentioned in
frequently, what other terms have been mentioned with the selected term, who
has mentioned the
term, latest tweets mentioning this term an icon to take action with, and the
like.
102701 Referring now to FIG. 37. a list page of a Social Media map platform is
displayed. In this
view,. information may be provided in the form of lists, such as lists of
influencers, conversation
leaders, key amtent, terms, and the like. Information about each list member
may include rime,
screen name, M score, followers, mentions, follower/following status, and the
like. Lists may be
:sorted/filtered by any of the techniques mentioned. herein including by
influence, Ivi...Score (such
as with a slider or other user input), and the like. Users may take action
from the list view.
102711 In further embodiments, an analytical framework. for a coordinated
campaign
identification includes proposing a framework for analyzing fabricated social
movements. in
many embodiments, not only is them the ability-to:distinguish these movements
from truly organic
ones, there is also the ability to create a formal method for studying
patterns of fabricated, pseudo-
grassroots (also, "astroturf") collective action.
102721 it Will .be appreciated in light of the disclosure that any such
collective action may be
required to give the impression of a large group of pee* coalescing around a
movement that is
easy to describe and share with others. I f the group is not well-connected
enough,: then it may be
logistically difficult for any actor to organize the group's online behavior.
If the group is not
acting in temporal leekstep, then its message maynot achieve a high
frequency.. II embodiments,
low-frequency messageS do not appear as global trends; for example, Twitter's
"trending"
algorithm appears to identify topics that are popular now, rather than topics
that havebeert popular
fora while or on a -daily basis, to help you discover the hottest emerging
topics of discussion on
Twittertm. The many examples remain applicable to the myriad social platforms.
Finally, if the
group behind a fabricated social movement does not Oromote. it with a coherent
message, the
movement's impact on the general public may be blunted by conflicting
information.
102731 It will be appreciated in light-of the disclosure that these
constraints suggest a etaturat set
of three dimensions for.analyzing _fabricated social movements: I.)
the..semantic dimension (how
messages are formulated), 24 the network dimension (how accounts within the
campaigns are
connected to one another.) --And 3.) the temporal dimension (when messages
spread throughout the
66
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
campaign). In many embodiments, these dimensions, and their intersections,
yield discrete signals
that can be used to scrutinize social media operations and assess if they
display a Suspicious degree
of hidden coordination.
102741 In. embodiments, the. framework operates on three levels;Event, the
level of an entire
social campaign; 2.) Segment, the level of a community of users participating
in a social media
campaign (e.g., Russian social Media WI accounts),. and 3..) Actor, the level
of an individual user
participating in a social media. campaign.
02751 Table I below shows examples of the three-dimensional analysis -
framework in more detail
specifically, the signals relevant for particular. combinations or level and
dimension, ftvill be
appreciated in tight of the disclosure that not every combination of level and
...din-ten:40n has
corresponding relevant signals,
67
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
Network Temporal Semantic.
.Event: how concentrated is
online participation
movement? Does. it cover a
Segment; ::How does
-.broad range of politically /
.
socially I culturally distinct participation in the
movement vary between
communities;. or: is it
differenteommunities
contained innhoritogenconS
== = and overtime? Are
"echo chaMber"? Segment: How
particular communities
topically diverse is
Segment: do communities: always lagging behind all
. discourse among.
Network = = the. rest in participation
actors who participate in the comnitmities
(taking time to fOrmalate
IllOVetnent pay a res ? participating in .the
ponse)
disproportionate attention to movement?
each .other?
Actor: how long does
the average actor
Actor: do _actors who
participate in the movement participate in the
movement?
-do se- in conjunction with their
communities, or
independently of them?
Event: Does
participation in the
movement follow an
unusually temporally
regular pattern, when
compared to spontaneous event/Segment/Actor:
human posting behavior? How does the diversity
of' the discourse among
Segment: do all participants /
specific
communities of actors communities /
Temporal
coordinate their individual actors
activities, even across participating in the
time zones? movement vary over
time?
Actor: Do some actors
behave similarly to pre-
identified troll or
spambot accounts. with
regard to their temporal
posting patterns?
Event/Attar: How
topically diverse is the
.discourse around the
Semantic
movement among all
actors / individual
. actors?
Taible:1.Three-Dimensional Analysis 'Framework
68
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
102761 This framework is a helpful methodological tool, but it would not be
useful without
operational definitionsi, which are captured via mathematical metrics of -
campaign activity. In
embodiments, each signal in Table 1 above is mapped to a discrete metric in.
Table 2. Further
detail regarding key definitions for understanding these metrics, and any non-
obvious activity.
metrics are provided herein.
Table .1 Table
Table 1 Level Metric
Row Column
Network Network Event Entropy E
Network Network Segment Inter-community homophily
% of actor's community participating in
Network Network Actor campaign, by number of individuals
or
total posts
Time delta between peak date of
Network Temporal Segment
campaign participation by segment
Network Temporal Actor Commitment by actor M
Semantic :Diversity by Segment Omega
Network Semantic Segment
LIS
Temporal Temporal Event Campaign Peakedness- P
Dynamic Time Warp alignment between
"temporal Temporal Segment
Segments 1)5
Dynamic Time Warp alignment. between.
Temporal Temporal Actor
Users DU
Semantic Diversity over time by Event
Temporal Semantic Event/Segment/Actor
Segment / Actor airõ af S. tr A
Semantic Diversity by Event i Actor f/r,
Semantic Semantic Event/Segment/Actor
Table 2. Mapping of Signals to Metrics
Key Definitions
Network
(02771 In many embodiments, the network dimension assumes that actors
participating in a
campaign are connected to each other in a directed network G (i.e.. a
connection from user a to
user h does not imply the reverse). Twitter following networks are an example
of directed
69
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
networks: many people follow TwitterThl celebrities, but those celebrities do
not follow their fans
hack. as A general rule. Other social media platforms and 'connected platforms
are applicable.
Segment
1027.8] 'When calculating metrics at the network level, it is assumed that
each actor partkipating
in a campaign belongs to exactly=tine community c, where e represents a group
of actors with
similar interests, whether social, ptilitical,:Or otherwise,
identifying Networks and Communities
102791 in order to identify relevant networks and communities within those
networks,- network
segmentation technologies are leveraged such as hierarchical agglomerative
clustering. in many
examples, it may be shown that network' segmentation framework,. based on
hierarchical
-.agglomerative clustering has. been tested on more = than eight hundred
different sociocultural
contexts with many academic applications. By way of many examples. the .unit
of analysis is a
"map," which may be a =collection of key social_ media accounts around a
particular social context.
A map may be composed of "nodes," which are the social media accounts in
question. Each node
may be connected to.. one or more nodes in the map through "edges" and edges
may represent
'social relationships embedded in the respective social media platform (e.g.,
"following" tbr
Twittertm, FacebookTM. or the like).
102801 In embodiments, each node in the map may belong to exactly one.
"segment" and one
"group." By way of these examples, a segment may be a collection of nodes with
a shared pattern
of interests. (e.gõ a collection 01'1*We" accounts who all follow US Tea Party
politicians).
Each segment may have a label (e.g., "Tea Party"). A group may be a collection
of segments with
similar interest. profiles (e.g., .a collection of "Tea Party,"
"Constitutional Conservatives," etc.
segments into a "Conservative" group). The process for generating segments,
groups, labels, and
colors for a map be fully or partially automated, as follows: a proprietary
clustering algorithm may
automatically generate- segments and groups for a map: subsequently, the map-
making process
may use supervised machine learning, to generate labels tbr segments and
groups from human-
labeled examples. At the end of the automated process, a Subject Matter
Expert, an individual
well-versed in the topic and/or geographical area covered by the map, may
perform a quality
assurance check on the segment and group labels.
Key Metrics FAtplained
192811 TO illustrate metrics in this section, a toy campaign example may be
employed. The
example consists of 100 users connected in a network G. The. network G further
breaks down into
exactly. two communities A and B,. each with exactly one halfof the total
population. The overall
number of connections from members of A. to any other actor in the network. is
500, while the
number of connectioas from members of 4 to members of B is 200. The campaign
proceeds over
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
the course Of ten days, and the first of those days features the highest level
of campaign activity,
with exactly. one quarter of all actors participating.
Entropy E
10282] This metric is the degreeto which a particular campaign is concentrated
in one community
versus diffused among many different communities. Given a mapping of users to
Communities,
which is described, in more detail below, the entropy of a campaign may be, as
known in the art,
the information theoretic entropy of the distribution of users active in the
campaign among
different communities. In the toy example, the Entropy of the campaign may be:
id
E = p(c(0)log L,(c(i)) = ¨0.5/og2(0.5) 0.5iogz(0.5) =
In general,. it May be:Shown that low values of E represent campaigns
concentrated in one
community, while: high. values of E represent campaigns distributed among a
wide array of
communities,
Inter-community Homophily H
102831 It is known in the art that the inter-community Homophily 11 is the
degree to Which
communities active around the campaign are more interconnected than one
...would expect by
random -chance,. Mathematically, H is calculated for an ordered
pairofeommunities A, B. The
quantity HrA,B),IS theratio of the actual number of connections from
metribersof A to members
of B, E(A,14. Witmermalizing factor p that assumes that members ofillnake
theirootineetiOS-to
all other nodes at random. In the random. baseline, the number of connections
frommettibemor
A to members of.B is the number of all connections from members of A to any
other node in-the
network a-multiplied by the fraction of G that B represents. In the toy
example, the Homophily
from coinmunity A to community B is:
E (A, B) 200
H (A, B) = = 0.8
500 * 0.5
102841 Values of H blew IA) may be shown to represent heterophily, or lower-
than-expected
intereormectivity between coMMunities. Values of /I equal to 1.0 may he shown
to represent the
baseline random expectation. Values of H above 1.0 may be shown to
represenrhomophily, or
higher4han-expected interconneet&Ity,
102851 H is superlinear, so a value ot.k= 4.01Stnueb.more-than twice as
interconnected as H =
71
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
192861 While the random baseline for flomophily is established in the citation
above, it will be
appreciated in light of the disclosure that it may be an excessively low
baseline for such empirical
analyses. Therefore, when possible, H values are.used for community pairs
where there. may be
expected low high values (e.g., ideologically separate ideologically aligned
communities) in
the same networked terrain as the case study as a baseline.
Commitment M
102871 Commitment to a particular campaign is measured in two ways: I.) M, the-
number of
subsequent engagements with the campaign by an actor; or2.) Mr, the length of
tittiebetween first
and last recorded engagement with the campaign by an actor.
Semantic Diversity ca
102881 Semantic diversity of a particular actor's / segments / campaign's_
messaging is based on
the assignment of messages 10 topics. As known in the art. LOA is a common
method for
identifying topics in text data. Once messages have been assigned to topics, a
semantic diversity
score. may be calculated for the message set. The- authors of the referenced
work may represent
their measure of semantic diversity as the probability that: two documents
chosen from the corpus
at random with replacement will be on the same topic. By way of these
examples, the corpus may
be the message set, and the documents may be..user Tweet histories, post
histories, etc, aggregated
.by user. In many examples, the LOA algorithm may run for 15 iterations, with -
a nulriber of topics
no less than 20% of the number of documents and no trim than 30. iterations
and may average
semantic diversity over 20 distinct runs of the LOA algorithm, on the same
corpus to smooth out
variations due to the initial conditions for a particular run.. For topics
that dO not co-Occur in
documetits,a topiemay be assigned a distance-score ofõ1.000.
102891 In embodiments, versions of CI are run for individual users (i/a),
communities (0.c), or
entire aumpaigns-:0). These metrics can also be rtm for all messages within a.
particular time
.perioit('21.4)' to Otdetilatethechafige in semantic diversity over time.
102901 SentantiediVeraity scores of less than one may represent users who
exclusively post about
the same topic, -eharactetiatie of fabricated campaigns. Semantic diversity
scores between 1 and
100 may represent users who post on a variety of topics, characteristic of
normal human activity.
Finally, semantic diversity scores above 100 may represent Users who post on
an extremely
diverse set of topics; 'characteristic Of spainhots or users who bridge-
different Cultural -and/or
-110000:communities (e.g.., users who post in different languages, etc.)
Campaign Peakedness
102911 Campaign Peakedness may be defined as the fraction of all activity that
occurs in the day
with the most campaign-related activiv during some time frame. In the toy
example, P = %
0.25.
72
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
Dynamic Time Warp Alignment D*
102921 The Dynamic Time Warp is an algorithm known in the art for comparing
two temporal
sequences of activity. in the many embodiments, the Dynamic Time Warp may be
used to
compare the. activities of individual users :(Ptr):oreritire segments (D4
In.general, the Dynamic
Time Warp between two sequences $./ at4S2 is the number of warping
transformations that are
requiredtb,:ehange ,S1 into 2. In many examples, Dynamic Time Warp may be
used to identify
bots andtrOlis=in a different social media setting.
02931 In-inany examples, this framework and these metrics on eighteen case
studies of political
campaigns have been tested in seven differeat-sogiocalturat settings, spanning
three continents
and. Sbt YearS'in all. These StudieS included ten groups:. of Twitterlw
hashragS linked by subject
Matter expel/SOME) to known coordinated campaigns; and eight groups of
Tµvitterim hashtaus
linked by Sivift-tokriown spontaneous campaigns. Based on the eighteen case
studies, it may be
shown that clear differences between coordinated and spontaneous campaigns
across sociocultural
setting and time for four of the metrics listed above: Entropy Commitment by
subsequent
engagements Me, Time delta, and Peakedness P.. The same analysis alto showed
that at least one
especially coordinated campaign showed extremely low values of Semantic
Diversity by Event
fit and high Dynamic Time Warp alignment DSbetween the activity of different
segments.
1112941 In further embodiments, methods and systems are disclosed for
identifOng markers of
coordinated activity in social media movements that may identify a largo
number e=faccounts that
may be controlled by a small number of coordinated entities that may result in
a measurable lack
of diversity of a similar number of accounts controlled by uncoOrdinated
individual actors. To
facilitate the methods and systems of identifying markers of coordinated
activity in social media
=movements, a framework of signals (or metrics) along at least three
dimensions may he
constructed and may 'include, without limitation:
102951 A Network dimension that may, for example, represent how accounts are
connected;
192961 ATemporal dimension that may represeriµforexample, patterns of
messaging across time;
and
102911 A:Semantic dimension that may represent, for example, diversity of
topics and meaning.
102981 From this framework, a plurality of hypotheses may be derived for
"signals" exploring
'potentially hidden coordination on Social media movements on a social media
channel Such as
14itterThi, Facebook.rmor the like. The exploring potentially hidden
coordination on social media
movements on a social media channel may occur at the level of the entire
campaign (e.g., nine
signals), a cluster level of the campaign (e.g., a set of well interwoven
accounts), at the individual
account level; and the like. In embodiments, the plurality of hypotheses may
include twenty-five
or more such hypotheses. Empirical evidence associated with these signals can
be shown across
73
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
a number of case studies of known coordinated (i.e., inorgartie centrally-
controlled) and
spontaneous (i.e.õ organic, individually) campaigns. In embodiments, three of
the campaign
signals may systematically reveal coordination in social media movements on
TwitterTm,
Facebookrm and other platforms. Some signals, either at. the cluster or at the
individual account
level, may facilitate campaign analysis, and some of them may he transformed
into campaign
-
level signals.
102991 (µ-arnpaign Minster /User Each campaign may include a set of "seeds"
from a specified
timeframe that may be, for example, a hashtag, a sentenee shared in. posts, a
URL shared in posts,
or the like. In embodiments, clusters may be. communities of users active
within the campaign.
In embodiments, users may be defined by their individual accounts, defined by
their Twitterni
handle, Facebooktm identification defined by their user name on other social
media platforms, or
the like.
103001 Network Terrain ¨ Campaigns may occur in a specific context referred to
as the "network
terrain." in one example, 'it will be appreciated 14 the light of the
disclosure that the
#BlackLivesNelatter movement may be better analyzed within its "network
terrain," which displays
the US political conversation on Twit-teem, FacehóokTM or other relevant
social media platforms.
In a representativelllodel, social media platforms. like Twitteirm, Facebookr4
may constitute a
eyber-social "network terrain" formed by the relationships (such as following
in Twittefrm,
Facebooklm, or the like among actors. The structure of the network or social
media platform may
determine who and what may be visible to whom, and thus it may be the social
landscape on which
the struggle for influence may occur. The methods and systems may include
analyzing ease study
campaigns across specific network terrain maps in order to understand the
relationships between
participants and the patterns of campaign propagation across specific online
communities (e.g.,
clusters or clusters discovered using machine learning analySiofnetwork
relationships and the
like).
103011 Campaign versus Investigatory Signals ¨ Signals measured at the cluster
and individual
actor (user) levels may facilitate investigating the inner workings of
specific campaigns, building
a more qualitative understanding of how these campaigns unfolded, and helping
form campaign
level metrics among Other things.
103021 Case StudieSHTo date, the methods and systems may include testing
signals set on a set
of case studies arid.ek.ernplary campaigns.
SIGNAL .SUMMARY
103031 Exemplary Investigatory Signals ¨ The investigatory signals may operate
at the cluster or
at the individual level. The investigatory signals may facilitate building a
qualitative
understanding of the dynamics of a campaign. and may provide tools to build
campaign-level
74
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
signals:. IQ:indicates a signal operating at the cluster level, and [U]
indicates a signal is operating
atiho -user level.
.1410041 The following are exemplary priority signals:
103951 Concentration in Lead Cluster [CI;
10306] concentration via Entropy [C];
103071 Day - peakedness [C];
103081 Temporal coordination per cluSterlej;
103091 Temporal coordination per user
103101 Client diversity per cluster [C]; and
103111 Time delta between clusters [C],-
103121 Other signals include:
f03131 Commitment by user (14
103141 Commitment by cluster (0;
103151 Account creation date diversity for cluster IC];
103161 Ilomophily (C);
103171 Language mismatch [C];
1031.81 Russian language profile % [C];
183191 .% in cluster also active [C];
103201 14) of hits inovrncluster KJ;
103211 Account creation -datediversity.[C];.
103221 Semantic diversitY by.user for user tweetsTm (or other postings) [L];
f03231 Semantic diversity by time slice by cluster [C]; and
103241 Semantic diversity by time slice by user pl.
193251 in embodiments, a priority signal name is Concentration in Lead
Cluster.
103261 The concentration in lead cluster signal description - Large-scale
spontaneous campaigns
may be more likely to engage participants from a range of different clusters,
whereas coordinated
campaigns are typically highly concentrated in a specific cluster of the
network. or social media
platform. The concentration in lead cluster signal (metric) evaluates the
degree to Which an entire
campaign's activity is concentrated in a particular cluster of participants.
The concentration in
lead cluster signal.#.n.etric) may 'Measure by the fraction of all campaign
participants who are
members of the most tampaign-active cluster in the network terrain map.
103271 The range of score value range of the concentration in lead cluster
signal (metric) is zero
to 100%. In embodiments, the concentration in lead cluster signal (metric)
value is computed by
determining the value of the concentration of the fraction of a campaign's
participants that are
members of the most active community in the campaign. In an example including
a 3-community
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
map, if 50: participants are from community A., 23 from community B, and 25
from community
C, then the Value athe concentration in lead cluster sights] (metric) forthe
campaign on this map
equals 50%. In embodiments, possible value.s. of the concentration in
leadeluster signal (or metric)
may be between 0 (i.e., not concentrated) and 100% (i.e4 fully concentrated in
1 cluster).
103281 The Concenintion in lead cluster signal (or inetrie)..-may be
consistent across 4 set. of
campaigns,:Whiehniay Over a variety of geographies and dates, It will be
appreciated in light of.
the disclosure that coordinated campaigns, on average, may be shown to have
larger values of the
concentration in lead cluster signal (of metric) than those of spontaneous
campaigns, It will also
be appreciated in light of the disclosure that there may be some. overlap
between thecoordinated
and spontaneous ranges due at least in part tO a large number of sociocultural
Settingt and time
periods in the data sets.
103291 An exemplary average value of the concentration in lead -cluster signal
for coordinated
campaigns is 48%.
103301 An exemplary range :of values of the concentration in lead cluster
signal score for
coordinated campaigns is 20"4 to 89%. The range here is the full range between
the lowest value
and the highest Value for this -category in the campaign.
103311 An exemplary value of the standard deviation of the concentration in
lead cluster signal
for coordinated campaigns is 0.21.
103321 An. exemplary average value of the concentration in lead cluster signal
for spontaneous
(organic) campaigns is 22%.
103331 .An exemplary range of values of the concentration in lead cluster
signal score for
spontaneous campaigns is 9% to 50%.
103341 An exemplary value of the standard deviation of the concentration in
lead cluster signal
for spontaneous campaigns is 0.12.
103351 In embodiments, the performance of theconeeramtion,in lead cluster
signal (metric) may
be sensitive to. the. specific terrain map being used. because the signal
(metric) may be less
successful if the terrain map used only captures the active participants_ in a
campaign. The
concentration in lead cluster signal (metric) may be more successful when
capturing the broader
-terrain in which the campaign under scrutiny unfolds..
103361 The methods and systems described lierein:tdso include computing the
value of the
concentration in. lead cluster signal (or mettle) 'using actions rather than
users and may measure
what proportion .of the total actions (Tweetsm or the like) in the campaign
that came from the
most active community. This approach can be shown to be. reliable because
heavy posters (those
who Tweetm or the like) may -create skews in the measurements.
103371 In embodiments, a priority signal name is Concentration via Entropy.
76
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
103381 The concentration via entropy -signal description ¨ The concentration.
via entropy signal is
another .apprOach to measuring concentration that looks at how the
participants are distributed
among.the active. communities In the campaign .rather than simply looking at
bow many of them
belong to the most prevalent.community. The concentration via entropy signal
(metric) may be
shown to be a useful signal for knowing if more than :one community is driving
a coordinated
campaign, which could be missed relying oothe concentration in lead cluster
signal (metric) alone.
The eonc.entratiOn 'via entropy signal (metric) may calculate the
concentration = of distribution
among all clusterS. In embodiments, .coordinated campaigns generally tend to.
have values of the
concentration via entropy signal. (metric) that are less than 2Ø
103311 The concentration via entropy -tigtµal. value range ¨ Relatively higher
valueS of the
concentration via entropy signal (metric) reflect more even distributions of
participants between
the communities active in the campaign. The lowest score is zero (all
participants belong to the
same community). The highest score depends- on the number of communities
active in the map.
Because the highest. number of communities in an exemplary case study map may
'be 50, the
highest entropy value in this example would be four (assuming a perfectly even
distribution of
participants amongst the 5(1 communities).
103401 How the concentration via entropy signal is. computed ¨ The
concentration via entropy
signal (metric) may be. an entropy of the distribution of participants among
communities. In an
example:with a two-community map,_tbe value of the Concentration via Entropy
signal would be
1.0 when %participants are from community A, 50 participants are from
community Bõ and thus
the distribution W-ould be 03,0,5.
103411 Exemplary formula for the concentration via entropy signal (metric):
tel
E = E p(c(i)Noggc(i))
103421 In the fomiula, c(i) is the count of participants in the ith cluster
and p(c(I)) is the fraction
of all participants coming from the jib duster.
103431 in embodiments, the concentration via entropy signal (metric) is based
on a logarithmic
scale, so a small difference in entropy belies a large difference in the
unevenness of the underlying
distribution. It will be apt:we-dated in light of the disclosure that a very
rough rule of thumb is that
a difference of one point in the value of the concentration via, entropy
signal may be equivalent a
change in concentration by a factor of three, so a campaign With the
concentration via crittepy
signal equal to two is three times more concentrated. in a. few clusters than
a campaign with the
concentration via entropy signal that is equal to three.
103441 Analysis in case studies The concentration via entropy signal (metric)
can be. shown to
77
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
be consistent across campaigns despite the variety of geographies and dates.
It will be appreciated
in light of the disclosure that coordinated campaigns, on average, have a
lower concentration via
entropy signal.
103451 An exemplary average value of the concentration via entropy ,-signal
for coordinated
campaigns is 1.43.
103461 An exemplary average rangc of values of the concentration via entropy
signal for
coordinated campaigns is 0.46 to:2;19.
103471 An exemplary standard deviation of the value of the concentration via
entropy signal for
coordinated campaigns is 0.57.
103481 An exemplary average value of the concentration via entropy signal for
spontaneous
campaigns is 2.52.
103491 An exemplary average range of values of the concentration via entropy
signal for
spontaneous campaigns is 0.69 - 3.38.
103501 An exemplary standard deviation of the value of the concentration via
entropy signal for
spontaneous campaigns is 0.71.
103511 in embodiments, the concentration via entropy signal (metric) may be
useful to analyze
"battleground campaigns" where a few clusters fight for control over the
social media narrative,
e.g., on a dedicated hashtag, where these campaigns may be comentrate.d in
these few
communities and simply using a measure focused on the lead community may Miss
this activity.
103521 In embodiments, a priority signal name is DayPeakedness.
(03531 The daypeakedness signal description ¨ A coordinated campaign,
typically, may exhibit
sustained activity by the accounts promoting it. Spontaneous activity, in
contrast, is characterized
by "bursty" cascades of activity. In embodiments, the daypeakedness signal may
detail the
=percentage of all activity that the busiest day of the campaign mayrepresent.
W541 The daypeakedness signal (metric) of a campaign is measured as the
percentage of-
caMpaignactions (Tweets"' or the like) that take place on the most active day
of the campaign.
ltiiI he appreciated in light of the disclosure that generally spontaneous
campaigns appear to be
more "bursty" because, for example, spontaneous campaigns exhibit more of a
peak (or more of
nntnber of peaks) than coordinated campaigns.
103551 In erabodittents, the range Of the values of the daypeakedness signal
(metric) is 0% to
100%.
103561 in embodiments the. value of the daypeakedness signal (metric) is
computed by
determining ate-fraction .efall activity that occurs on the day with the most
campaign-related
activity. Examples inetudeaeampaign that proceeds over the course of ten days,
and the first of
those days ' .feattrieS 'the ifighe0 level of campaign activity, with one-
quarter of all actors
78
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
participating. In this example, the. value of the daypeakedness signal
(metric) is 25%.
103571 It will be appreciated in light of the disclosure that one-eighth of
all activity in coordinated
campaigns, on average, happens during peak day, whereas over one-third of all
activity for
spontaneous campaigns happens during peak day. In embodiments, the
daypeakedness signal
(metric) can be shown to be consistent across campaigns despite the variety of
geographies and
dates. Byway of this example, coordinated campaigns,. on average, may have, a
lower value of
the daypeakedness signal (metric) than spontaneous campaigns. It will be
appreciated in light of
the disclosure that there may be some overlap between the coordinated and
spontaneous ranges
due to the large number of.socincuiturai settings and timperiods in the
campaign.
103581 An exemplary average value Of the daypeakedness signal for coordinated
campaigns is
0.14.
103591 An exemplary range of values of the daypeakedness_ signal for
coordinated campaigns is
0.08 to 0.22.
103601 An exemplary standard deviation of the valued f the daypeakedness
signal tbr coordinated
campaigns is 0.05.
103611 An exemplary average value of the daypeakedness signal tbr spontaneous
campaigns is
0A I
103621 An exemplary average, range of values of the daypeakedness signal for
spontaneous
campaigns is 0 to 0.71.
103631 An exemplary standard deviation of the value of the daypeakedness
signal fbr spontaneous
campaigns is 0.21.
103641 The daypeakedness signal. (metric) may be sensitive to date-
boundaryltime.zones most
notably when the campaign is being analyzed only over the last few days. in
embodiments, the
sensitivity of The daypeakedness signal (metric) may be improved by allowing
it to be less
sensitive to time zones.
103651 It will be appreciated in light of the disclosure that there. are other
possibly more complex
ways to calculate the value of the daypeakedness signal. In embodiments, the
peak time may be
identified as. the median of time stamps of a dynamic phenomenon to be able:
to observe a
logarithmic distribution of volume, around the peak. The methods and systems
described herein
may identify peak -a as days when. volume exceeds two standard-deviations
above the median, and
may calculate the value .of the daypeakedness signal as a fraction of all
content that occurred
during a 24-hour period. It will be appreciated in light of the. disclosure
that the median volume
may be used instead of mean volume due in part to the:observation that volume
follows a skewed
distribution, so the mean may not be an appropriate-statistic to use to
characterize it. The measure
of peakedness in the methods and systems described herein may bc relatively
less sophisticated
79
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
and, therefore, may be easier to interpret while giving a good initial
itnpr&ssion of the utility of
the signal from a social media platform for identifying coordinated campaigns.
103661 In embodiments, the value of the daypeakedness signal (metric) may be
affected by the
overall time range of a-campaign. .By way of this example, if .a campaign
lasts three days, then
the value of the daypeakedness signal may not go below .÷% but if the campaign
lasts 10 days,
then the value-of The daypeakedness signal cannot go below 10%. IO
embodiments, -campaigns
may last as little as one week and may last as long as several months. The
value of the
daypeakedness signal may be shown 16 ibllOw the pattern described in the
campaign value
examples across these time ranges..
103611 In embodiments, a signal name ISCommitment: Average-.Posts Count in the
Campaign.
193681 The commitment: average posts count in campaign signal description -
Campaigns
typically feature numerous -die-hard supporters who post repeatedly and fewer
casual participants
who merely chime in. This commitment: average posts count in campaign signal
(metric) may
capture the degree to which a campaign's body of actors sticks with -further
posting after. their l'irst
engagement with the social media Platform. In embodiments, the value of the
commitment
average pests count in campaign Signal (metric) can include the average number
of campaign-
related posts that participants publish after their first campaign post.
103691 The range of values of the commitment: average posts count in .campaign
signal (metric)
is bounded by the lowest value being zero which corresponds to a user only
posting once about
the campaign. In embodiments, the commitment: average posts count in campaign
signal (metric)
may have a range of values between 0 And 10 posts. it will be appreciated. in
light of the disclosure
that the maximum value of the commitment: -average posts count in: campaign
.signal (metric)
could be much higher. In one example, participants in a campaign. may be very
dedicated and
may post 100 times about a certain subject during the scope of analysis, and
the. like.
(03701 To compute the value of thecottimitrrient: average postsecaunt in
campaign signal (metric),
the Methods and systems disclosed herein determine, the average number of
subsegnotit
participation actions, e.g., Tweetsm (or other posting) with campaign hashtag,
across all
participants in a campaign. In embodiments, participants (i.e., posters) in a
campaign can be a
smaller subset of participants in a Map. In embodiments, the map may capture
.some of their
-followers and/or other members Of the network terrain when thoseate highly
connected to active
participants in the campaign in order to compute the commitment: average posts
Count in
campaign- signal (metric), only participants who actually posted about the
campaign are taken into
account. For example, when aparticipant posted through Twitterlw, FacebookTM,
or the like with
a campaign-related hashtag twice, their commitment is 1,0. In embodiments,
campaign
participation can include Tweetsm or the like with campaign-related hashtags
(for campaigns
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
organized around a hashtag). Tweets or the like with links to a Video or
article (tbr campaigns
organized around a video or article), retweets of the above tweets and the
like. Examples of out
of scope for participation include favorites of tweets with campaign-related
hashtags or links or
(liprreplies or *mentions of Tweetsrm (or the like) with campaign-related
haShtags or links.
103711 It will he appreciated in light of the disclosure that participants in
spontaneous campaigns
post more about their campaigns than participants in coordinated campaigns;
It. will also be
appreciated in light of the disclosure that this pattern may be
counterintuitive, as one may expect
participants in coordinated campaigns to be extrinsically motivated to hit
certain participation
targets, (e.g., by being paid by number of posts), and thus to post more than
participants in
spontaneous campaigns, who lack such motivation.
103721 An exemplary average value of the commitment: average posts count in
campaign signal
(mettle) .for coordinated campaigns is 2.52.
103731 An exemplary average range of values of the commitment: average posts
count in
campaign signal (metric) for coordinated campaigns is 1.28 to 3.40.
103741 An exemplary standard deviation of the value of the commitment: average
posts count in
campaign signal (metric) for coordinated campaigns is 0.84.
103751 An exemplary average value of the commitment: average posts = count in
campaign signal
(metric) for spontaneous campaigns is 3.53.
103761 An exemplary average range of values of the commitment: average posts
count in
campaign signal (metric) for spontaneous campaigns is 1.39 to 6.07.
103771 An exemplary standard deviation of the value of the commitment: average
posts count in
campaign signal (metric)for spontaneous. campaigns is. 1.48.
103781 In embodiments, the commitment: average posts count in campaign signal
.(metric) can he
analyzed at the community level, at a cluster level, and a -participant level.
The commitment:
average posts count in campaign signal (metric) can be analyzed at the
community 1Oct to single
out communities with participants being particularly committed to a campaign.
The commitment
average posts count in campaign signal (metric) can be analyzed at the
participant level to
represent individuals who have extremely high commitment values, e.g., posting
about a campaign
one hundred times.
103791 In embodirtients, the -comthitment: average posts count: in campaign
signal (metric) is
focused on participations alter the first post and complemented by a
measurement of the
proportion of participants in the campaign who have only participated once.
103801 In embodiments, the commitment: average_ postscount in campaign signal
(metric) may
be combined with a commitment: average time range of participation. signal
(metric) into a
commitment: post regularity signal (metric) that may capture the deviation of
campaign
81
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
participants from natural human attention patterns.
103811 In embodiments, other statistical properties Of the:distribution
orposts per user may be
part of refining the commitment metrics In embodiments, there. may be. a
natural shape of this
distribution for spontaneous campaigns and that natural shape may be skewed.
it will be
appreciated in light of the disclosure that the commitment: average posts
count in campaign signal
(metric) may Make average Oa count an inappropriate metric in Many long
daratiOn situations:
Instead, it may be possible to be able to identilY coordinated campaigns by a
lack of skewness
and/or the presence of a second moment at some value above one, which may both
be indicative
of an unusually large perventage of participants posting multiple times about
a campaign, e.g., due
to a coordinating body paying these participants per post'.
103821 In embodiments, the commitment: average posts count in campaign signal
(metric) may
be normalized to take into accotintavenige posts per users in order to control
for users with a very
heavy activity across all campaigns:
103831 in embodiments, a õpriority signal. parte JO Commitment: AVentge Time
Range Of
Participation.
103841 The COMITOTICilt: average time range of participation signal
description ¨ in the desire to
determine whether participants in this campaign. are die-hard supporters
orjust people who chime
in,, the commitment: average time range of participation signal (metric) may
be used to facilitate
looking at bow long (in days) participants remained engaged in pushing the
campaign. in
embodiments, the loyalty of participants to the campaign may be measured by
time range (in days)
for their Campaign-related Tweets .1" (or other postings) that may be averaged
across all
participants.
103851 The range of the values of the commitment: average, time range of
participation signal
(metric)is an unbounded value and therefore can be zero days to the Wad length
of the campaign,
103861 In embodiments, the commitment: average time range of pattielpatiOn
Signal onovio may
look at the time frame between first and last participation action that can be
averaged across all
participants in a campaign. By way of this example, the 'commitment: average
time range of
participation signal (metric) may measure whether actors participate in a "one-
oll" way (one
TweetTm and done) or demonstrate a commitment to the campaign (multiple
Tweetsw or Other
'postings over time).
103871 it will be Appreciated in light of the disclosure that participants in
coordinated campaigns
engage with the campaign over a longer period than participants in spontaneous
campaigns. It
will also be appreciated in light of the disclosure that participants in
coordinated campaigns may
be more likely than participants in spontaneous campaigns to receive
extrinsic' motivation, such
as payment, for engagingwith the campaign and, as such, the extrinsic
motivation may lead to a
82
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
longer engagement period than intrinsic motivation.
(03881 An exemplary average value of the commitment: average time range of
participation signal
(metric) for coordinated campaigns is 7,24.
103891 An exemplary average, range of values of the commitment: average time
range of
participation signal (metric) signal for coordinated campaigns is 0.08- to
22,13 days.
103901 An. exemplary standard deviation of the value of thecommitment average
time range of
participation signal (metric) for coordinated campaigns is 9.04 days.
103911 An exemplary average valuta the commitment: average time range of
participation signal
(metric) for spontaneous campaigns is 1.53 days.
103921 An exemplary average range of values or the committnent average time
range of
participation signal (metric) for spontaneous campaigns is 0 to 3.36 days.
103931 An exemplary standard deviation of the value of the commitment: average
time range of
participation signal (metric) for spontaneous campaigns is 1.21 days..
(03941 it will heappreciated in light of the disclosure that the
cotrirnitnient: average time ranee of
participation Signal (metric) may he affected by the overall time.ran.geof a
campaign, e.g., if a
campaign lasts three days, then this metric cannot go above a value of three.
In embodiments, the
commitment: average time range of participation signal (metric) may be
combined into a
commitment: post regularity signal. that may capture. the deviation of
campaign participants from
natural human attention patterns.,
(03951 In embodiments, a signal name is Semantic _Diversity for all Messages.
(03961 The semantic diversity for all messages signal (Metric) description ¨
The semantic
diversity for all messages signal (metric) looks to detail how generally on-
message is the
campaign. The semantic diversity for all messages signal (metric) also looks
to determine whether
the interaction or activity appears like .a diverse conversation covering a
range of topics and
expressidtis or may be a- fairly uniform campaign with low semantic:
diversity. it will be
appreciated in light: of the disclosure that people tend to TWeefru (or
otherwise post) on a variety
of topics related to their daily lives, work, and interests. A group trying to
promote a coordinated
campaign, however, may be interested only in the narrow range of topics
relevant to that
campaign, in embodiments, bets or propaganda 'accounts may also be interested
in any Tweet"'
(or applicable posting) relevant to any campaign they -are trying to push, and
therefore could be
Tweetingx" (or otherwise posting) on art extremely wide range.Orippics.- In
embodiments, the
semantic diversity for all messages signal (metric) may be measuring the
extent to which
participants in the campaign are Tweeting m (or otherwise posting) on an
intermediate range of
topics, which suggests that their activities are spontaneous and human rather
than automated or
coordinated to propagate a Specific message.
83
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
10397j In embodiments, the range of values of the semantic diversity for all
messages signal
(metric) is zero to 100%.
103981 In embodiments, raw values of the semantic diversity for-all messages
signal (metric) fall
into three categories: (i) When the value of the semantic diversity for all
messages signal (metric)
is <1 (less than one), then it may represent users who exclusively post about
the same topic, which
may be a characteristic of fabricated campaigns. (ii) When the value of-the
semantic diversity.for
all messages signat:(mettic) is between one and 100, then it may represent
users who post on a
variety Of topics and-being.tharatteriStie Of normal human activity. (iii)
When the value of the
semantic-diversity for all messages signal (metric) is above 400, then it may
represent users who
post on an extremely diverse set atopia,::characteristicasparnbotio*users who
bridge different
cultural and/or linguistic communities (e.g., users who post in different
languages,. etc.:). In
.embodiments, the semantic diversity for all messages signal (metric) may be
set to be bounded at
1000 because it may be necessary to fix a maximum value for the "distance"
between any pair of
topics, for which no document includes terms, from both topics. It. will be
appreciated in light of
the disclo.sprothat mathematically the distance should be infinity but,
typically, it can be to set the
value to 1000. The percentage of users with the semantic diversity for all
messages signal (metric)
may-be greater than or equal to 1.0 and less than 100 and thus varies between
zero and .100%.
103991 Ilow the semantic diversity for all messages signal (metric) is
computed ¨ The value of
the semantic diversity for all messages signal (metric) of a particular
actor's (or cluster's, or
campaign's) messaging may be based on the assignment .of messages to topics.
In embodiments,
the computation of the semantic diversity for all messages signal (metric) may
use a Latent
Dirichlet Allocation algorithm. By way of this example, once. messages have
been assigned to
topics, the semantic diversity for all messages signal (metric) is determined
for the message set.
In embodiments,. the measure of the value of the semantic diversity for all
messages signal (metric)
is determined as the probability- that two documents Chosen from the corpus at
random with
replacement will be on the same topic.
104001 In the current exemplary case, the corpus is the message set, and the
documents may. be
user Tweeftm (or other posting) histories, aggregated by user. The Latent
Dirichlet Allocation
(WA) algorithm may be.:run for fifteen iterations with a number of topics no
less than 20% of the
'number of documents and no more than 30%. An. average value of the semantic
diversity for all
messages signal (metric) over twenty: distinct:rims of the WA. algorithm is
used on the same
:corpus to smooth out variations due to the initial conditions for a
particular run. In embodhuents,
a topic distance score of 1000 may be assigned to the semanticdiversity for
all messages signal
(metric) for topics that do not co-occur in documents.
104011 Because the. focus of the many embodiments is differentiating
coordinated and/or
84
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
automated campaigns from spontaneous and human-driven campaigns, the semantic
diversity for
all messages signal (metric) as the percentage of all users in a. campaign is
computed with raw
diversity score falling into the, range of normal human activity, i.e., the.
metric being -greater or
equal to 1.0 but less than 100. In embodiments, the semantic diversity for all
messages signal
(metric) may refer to all campaign-related messages,
104021 The values below show theipmentage of users with the semantic diversity
for all messages
signal (metric) greater than or equal to 1.0 and less than 100Ø
104031 An-exemplary average value of the semantic diversity. for all messages
signal (metric) for
coordinated iCampaims is 55%,
104041 An exemplar*-average range of values of the semantic 'diversity fOr all
messages signal
(metric) for coordinated campaigns is 17% to 90%.
104051 An exemplary standard deviation of the value of the semantic diversity
for all messages
signal (metric) for coordinated campaigns is 36.59%.
10404) An exemplary average value of the semantic diversity for all messages
signal (Metric) for
spontaneous campaigns is 71.3%.
1040.7.1 An exemplary average range of Values of the semantic diversity for
all messages signal
-(metric) for spontaneous campaigns is 50% to .98%.
ROM An exemplary standard deviation of the value of the semantic. diversity
for all messages
Aignal(inetric) for spontaneous campaigns is 21.2%.
104091 In embodiments, the semantic diversity for all messages signal (metric)
may be very
sensitive to cottfOttnds, By way of this example, flews organizations may tend
to have low
semantic diversity because news organizations may post the same story
headlines over and over
even though sit& news organizations are not coordinated actors. Moreover,
Tweetsm (Or other
postings) in one language tend to be more coordinated than Tweets", (or other
postings) in
multiple languages, because the Latent Dirichlet Allocation (LDA) algorithm
may not translate
terms across languages.
104101 At the same time, the semantic diversity for all messages signal
(metric) may point to the
differentiation between natural language use and the use of language to push a
particular message.
It will be appreciated in light of the diSelosurethat coordination around a
message may require
that that Message May be. as clear and simple as possible, whereas natural
language can he
complex, Metaphorical, and even slightly confusing. To that end, coordinated
campaigns may,
therefore, not wish to increase the -semantic diversity of their messages even
if the technical or
organizational opportunity was available.
104111 In embodiments, the semantic diversity for all messages signal (metric)
includes separating
language diversity from semantic diversity either by grouping TweetsThl (or
other postings) by
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
post language prior to analysis or using automated machine translation to pre-
convert all TweetSA'm
(or other postings) to the same language. The semantic diversity for all
messages signal (metric)
also includes leveraging existing natural language processing approaches to.
identify certain kinds
of low-semantic diversity language that may not be of interest, e.gõ. news
headlines and press
releases.
104121 In embodiments, the semantic diversity for all messages signal (metric)
may measure the
temporal alignment of campaign-related Tweets", (or other postings) for all
participants. It will
be appreciated in light of the disclosure: that users generally do not time
their tweets tor other
postings) to coincide with the. Tweets7g :(or postings):iof others. When the
Tweet-1m -(cn Other
posting) histories of Campaign participants follow The Same pattern of ebb and
flow, especially
across time zone boundaries, this may be. evidence that an actor is
coordinating the activities of
participants to create a concentrated temporal burst of engagement. The
semantic diversity fin all
messages signal (metric) may include temporal coordination of TweetsT" (or
other postings)
between campaign participants measured by alignment of Tweetim (or other
posting) histories
across all participants in the campaign.
104131 in embodiments, the range of the values of the semantic diversity for
all messages signal
(metric) is between 0% and. 100% and represents the percent alignment of two
users` temporal
normalized sequences of participation in the.campaign. Toward that end; 0%
alignment may mean
that the users' sequences do not match at all, while 100% alignment may
indicate a perfect match.
104141 In embodiments, the semantic diversity for all messages signal (metric)
may be computed
with a dynamic time warp algorithm for comparing two temporal sequences of
activity. in general,
the dynamic time warp algorithm between two sequences SI and 52 is the number
of warping
transformations that are required to change Si into S2. The methods and
systems described herein
may, _fOr example, use the dynamic time warp algorithm to identify bots and
trolls in a different
soda! media Setting. The number of warping transformations may be normalized
by the length Of
both sequences St and 52 and multiplied by 100 to get a percent value.
Finally, the normalized
number may be subtracted from 100 in order to calculate the percent alignment
of Si and 52.
104151 In embodiments, a priority signal name is temporal coordination
percluster.
104.161 The temporal coordination per cluster signal (metric) description ---
The temporal
coordination per cluster signal (metric) may look at the communities *ho
participate in this
campaign to identify different communities exhibiting very similar,pattems of
engagement that
may be considered as being -odd. In embodiments, the pattern of' the temporal
coordination per
cluster signal (metric) may he even odder when postings exist. indifferent
time zones. The
temporal coordination per cluster signal (metric). is measuring the temporal.
alignment of
campaign-related Tweetsrm (or other postings) aggregated at the cluster level.
With that in mind,
86
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
communities generally de not time their TweetsTm (or other postings) to
coincide with the
Tweetsm (or other postings) of other communities. When the Tweefrm (or other
posting) histories
of participating clusters fellow, the same pattern of ebb and. flow,
especially across time zone
boundaries, this may be evidence that an actor is coordinating the activities
of participants to create
a concentrated temporal burst of engagement.
104171 The range of Values for the temporal toordinatiOn per duster signal
(metric) is zero. percent
to 100%. The value of the temporal coordination per cluster signal (metric)'
represents the percent
alignment of two users' temporal normalized sequencesof participation in the
campaign. Toward
that end, 0% alignment may mean that the users' sequences do not match at all,
while 100%
alignment indicates a perfect match.
104181 The temporal coordination per cluster signal (metric) description ¨ The
temporal
coordination per cluster signal (metric) is_ a per-user take on examining
temporal coordination,
which might be helpful when other metrics are noisy. Temporal coordination per
user is
teehnically the emporal coordination between pairs of users. In embodiments,
the temporal
coordination per cluster signal (metric) may measure the temporal alignment of
campaign-related
Tweets'im (Or other postings) between individual campaign participants. AS
noted before, users
generally do not time their TweetsTm (or other postings) to coincide with the
tweets of others.
When the Tweet' (or other posting) histories of campaign participants follow
the same pattern
of ebb and flow, especially across time zone boundaries, this may be evidence
that an actor is
coordinating the activities of participan0o.create a concentrated temporal
burst of engagement.
(041191 The temporal coordination per Cluster signal (metric), especially its
heatmap visualization,
may provide a good high-level description of the rete of unusual coordination
across the users
participating in a campaign: ,Theitemporal. coordination per duster signal
(metric), however, may
suffer from the same overestiMutlinvof actual temporal coordination so the
algorithm may be
adjustable- for including iritheOtlettlatiOn the 'average temporal
coordination across users.
104201 In embodiments, a signal namels. client diversity per cluster.
104211 The client diversity per cluster signal (metric) description ¨ The
client diversity per cluster
signal (metric) may determine how accounts in a given cluster use Twittertm,
Facebookrm, or
other social media platforms. The client diversity per cluster signal (metric)
may also determine
hoW:Twitteem Users (or other posters or various relevant platfOrnis)golhrough
a mobile device,
a computer, or directly access. APIs of Twitter"' 'to Tweetrm or other social
media postings). In
one example, some clients may be used to coordinated Tweets rm (or other
social media postings)
and the client diversity per cluster signal (metric) may be used to determine
how coordinate are
the Tweets"' (or other social media postings), and are such coordinating
TweetsTm (-or other social
media postings) those that are: used heavily in some of the communities who
participate in this
87
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
campaign. It will be-appreciated in fight of the disclosure that client
diversity per cluster signal
(metric) is the same as the client diversity at campaign scale signal (metric)
but analyzed at the
cluster level.
104221 Ihere is no specific range of values applicable.* the client diversity
per duster sujnal
(metric) because it is a qualitative signal (metric).
104231 The value of the client: diversity per CluSter signal (metric) is
computed by using the
"source" field of the IweetTM (or other posting) to identify the client used
to make theTweet"'
(or other posting), as in the Client diversity at campaign scale signal
(metrie). ThevetheTWeetsT"
(or other postings) are-aggregated into clusters of the author of the Tweet' m
(orofheeposting) in
the campaign map.
104241 In embodiments, a signal name is Time Delta between Communities.
104251 The time delta between communities signal (metric) description ¨ the
time delta between
communities signal (metric) may identify a community that is engaging with the
campaign
significantly ahead of others In one example, this is due tokick-starting that
campaign or being
significantly behind maybe becausethere is a need to coordinate talking points
before engaging.
It will be appreciated in light of the disclosure that the time delta between
communitieS Menai
(metric) was -inspired by qualitative analysis initially done In the Syrian
Civil War Context such
that communities pretending to portray civilians while being led by military
intelligence engaged
with popular topics with a lag of several hours to days. Toward that end, the
time delta between
communities signal (metric) may examine when clusters are most active in the
campaign. By way
of this exalt*, the time delta between communities signal (metric) may measure
the distance
between a given cluster's, peak and the more-general peak of the overall:
campaign.
104261 In embodiments, the range of values of the time delta between
communities signal (metric)
represents a number of days. Negative valets may indicate that a community's
peak. of temporal
actiVity'happens before the average peak date for all other coMmunities.
IPositive values may
indicate the peak happens after the averagepeak date for all other
communities. ASCOre- of zero
may indicate a community peaking in sync with the rest of the communities.
104271 How the time delta between communities signal (metric ) is computed ¨
This metric
measures the number of days between the peak date of campaign participation in
a given cluster
and peak date of campaign participation averaged across all other Clusters. In
one example with
three clusters, where activity in cluster A peaks on 2-5 January 2017,
activity in cluster B peaks on
26 January 2017,- and activity in cluster C peaks on 27 January 2017, the
value of the time delta
between communities signal (metric) fbr A equals, -1.5, the value of the time
delta between
communities signal (metric) for B equals zero, and the value of the time delta
'between
communities signal (metric) for C equals 1 .5.
88
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
104281 in embodiments, the time delta between communities signal (metric) may
be helpftil to
analyze disputed hashtags, with both spontaneous and coordinated clusters.
engaging in the same
campaign. In embodiments, the time delta between communities signal (metric)
may point to the
natural logistical costof coordinating a message of a campaign in response to
a sudden event, such
as a late-breaking news story. It will be appreciated in light of the
disclosure that even the most
sophisticated coordinated campaigns cannot 'anticipate such events and at the
same time, they
cannot respond to these events spontaneously as it may distract from their
message and may hurt
the overall aim ofthe campaign. it will also be appreciated in light of the
disclosure that all
comdinatedicampsigns will need at least a little time to respond to late-
breaking events, and their
responSes will measurably lag behind spontaneous human reactions to the Same.
in embodiments,
the time delta between communities signal (metric) may include automatic
identification of
sudden events as they happen, e.g., by matching campaign-related terms against
Googleml News,
other news sources, and the like. A subsequent step may be to automatically
track responses to
the same events from campaign compared-to non-campaign-related clusters.
104291 in embodiments, a signatnarne is. c,Ottimitment by User.
104301 The commitment by user Signal (metric) description ¨ Loyalty of
participants to the
campaign may be measured by the number of times the. participants Tweet ro:
(or otherwise post)
about ;the campaign and time range (in days) for their campaign-related
Weetirm (or other
postings). The commitment by user signal (metric) may be measured by the user,
In
embodiments, the commitment by user signal (inottle). Wks at whether-
individual users are
particularly committed to a campaign, In embodiments, the commitment by user
signal (Metric)
may facilitate looking at users and their own commitments by determining
whether there are, for
example, people who Tweet.'" (or otherwise post) exactly 100 times, or some
predictable
predetermined amount. The value of the commitment by user signal (metric) may
facilitate
identifying, and singling out accounts that might be incentivized to
participate x number of times
or for x days straight.
104311 The ranee of values of the commitment by user signal (metric) an,
unbounded values
starting at zero, i.e., no subsequent actions,. zero days pass between -,first
and last action. In
embodiments, values for the commitment by user signal (metric) by Subsequent
actions are
between Zero and ten actions, thOse for commitment by time frame are between
zero And thirty
days.
104321 In embodiments, there may be users whose commitment by user .signal
(metric) is
extremely high and such behavior may also contribute to higher values
associated with the
Commitment: average time range of participation signal (metric) noted above.
104331 In embodiments, a signal name is Commitment by Cluster.
89
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
194341 Theeommitment by cluster signal (metric) description The commitment by
cluster signal
(metric) may be used. to determine whether a ispecific: cluster is
particularly committed to a
campaign. In embodiments; the commitment by cluster signal (metric) may
facilitate looking at
clusters and their own commitment& By. way of this example, the commitment by
cluster signal
(metric) may ftseilitate the determination of whether there are clusters that
Tweeerm (or otherwise
post) exactlie1:00:tit In embodiments, the commitment by cluster signal
(metric) may be used
to single outelutters-that might be incentivized to participate a certain
number of times or for a
certain length Of timc. Toone example, the commitment by cluster signal
(metric) may be used to
determine whether a group of accounts showed up.,--Tweeted114 (or otherwise
posted) .100 limes
osier five days, and .then left.
104351 In embodiments, the commitment by cluster signal (metric) may look at
the loyalty of
participants tO the campaign that may be measured by the number of times the
participants
lweetTM (or otherwise post) about the campaign and time range (in days). for
their campaign-
related TweetsTm (or other postings). in embodiments, the.
compaitmeetbyeltister signal (metric)
may measure the degree to which a body of actors in the campairtstiele with it
after their first
engagement with the campaign. It will be appreciated in light of the,
disclosure that the value of
the commitment by cluster signal (metric) for most humeri activity is- a
skewed distribution in
measurable contrast to coordinated activity that may include, those.:who
participate once with a
few die-hard supporters that participate a let Deviations from the skewed
distribution detailing
human activity may, there-fere, may reveal coordination. By way of this
example, if an actor
participates in a campaign exactly 100 times, this may suggest that they were
incentiViZed by a
_coordinating body to meet that threshold,
104361 The:range of the values of the commitment by cluster signal (metric)are
unbounded values
starting at zero, i.e., no .subsequent actions, zero days pass between first-
and last action. In
embodimeots, the value of the. commitment by cluster signal (metric) by
subsequent actions is
between zero and ten actions. In further embodiments, the value of the
commitment by cluster
signal (metric) by time frame is between zero and. thirty days.
104371 How the value of the commitment by cluster signal (metric.) is computed
¨ There are two
commitment metrics: (i)- counting the number of subsequent participation
"actions" (i.e.; Tweetsr"
or other postings with a campaign heshtae.9, and(ii) the tittle frame (in
days, can be fractional)
between first and last participation action. Both metrics may be averaged
across all participants
in a campaign. Both metrics may measure whether actors participate in a "one-
oft' way (i.e., one
Tweetm or other posting and done) or may demonstrate a commitment to the.--
eampeign
multiple Tweets"' or other postings over time).
104381 in embodiments, a signal name is Account Creation Date Diversity for
Cluster.
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
104391 The account creation date diversity for cluster signal. (metric)
description ¨ this signal
(metric) may facilitate observing how close in time all accounts participating
in a campaign were
created. 1190% of participatingaecounts within a given cluster were created
within a span of five
days, for example, then such activity may indicate a heavy coordination within
that cluster. The
account creation date diversity for cluster signal (metric) may be
particularly helpful to spot bats,
troll farms, and the like on networks using fake accounts generated in bulk..
104401 The range of Values of the account creation date diversity for cluster
signal (metric) is zero
to 4,015 days.. It will be appreciated in light of the disclosure that the
maximum range may range
from zero to the total day since the founding .of TWitteirm or the other
applicable social media
platforms: The vidties of the account creation date diversity for cluster
signal (Metric) in datasets
evaluated have included a rangeof zero to 1,200 days.
104411 How the account creation date diversity for clustersignal
(metric)iitomputed ¨ Account
creation date diversity for a particular cluster and campaign.combination is
the standard deviation
(in days) of Twitterim (or other applicable social media platform) account
creation dates for all
accounts in that cluster who engaged:with the campaign in question. As a
baseline, embodiments
may compare account creation date diversity for a particular cluster to
account creation date
diversity for the entire campaign.
104421 In .embodiments, a signalname is Homophily.
10443] The homophily signal (metric) description This signal (metric) may
facilitate looking for
communities that pay a "disproporti~ amount of attention. to one another, for
instance across
ideologies, language, culture, or the like, In embodiments, the homophily
signal (metric) can
identify disproportionate attention relationships between clusters measured by
a number of
following relationships between clusters. When looking at. communities
(clusters), it will be
appreciated in light of the disclosure that it is just .as important to
understand-who the community
pays attention to as who is in the community. With this it mind, the
.toittophiiy signal (metric)
may measure deviations from expected patterns of attention in social media. By
way of this
example, it will be appreciated in light of the disclosure that most people
may pay most of their
_attention to like-minded friends-and the vast majority of people may pay most
of their attention to
.friends in the same cultural and linguistic environment or in their affinity
in further examples,
the homophily signal (mettle) may facilitate the identification of patterns of
intense inter-attention
across ideologies, cultum and language that may imply evidence for
coordination.
104441 The range of values of the homophi ly signal (metrie)canbe shown to be
zero to ten.
104451 How the homophily signal (metric) is computed --- The homophily signal
(metric) as a
telltale of cluster attention is a ratio of the actual number of edges
connecting members of the
clusters compared to what would be expected under conditions where each
cluster paid attention
91
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
to every other cluster strictly in proportion to the clusters:size. Typically,
the baseline for such a
Signal (metric) in is random connection patterns. In embodiments, the
honnophily signal (metric)
includes relatively more aggressive baselines because no actual human
relationships follow a
random pattern.
104461 In embodiments, a signal name is Language Mismatch.
j0447) :The 'language mismatch signal (metric) description ¨ The default
language for a new
TWitteirm (or other social media) account appears to be English. Users may,
however, choose to
:ehang0..their profile language if they want. It will be appreciated in light
of the disclosure that
users ;posting frequently in a language that differs from their decal*
TwitterTm (or other social
media) profile language may be part of a foreign-language propaganda operation
on behalf of
some coordinated entity.
f04481 The language mismatch signal (metric) may measure the percentage of a
campaign's
TweetsTm (or other postings) - at both the cluster and campaign level - that
is in a language that
differs from the users' default TWitterm (or other social media) profile
language.
104491 The range of values of the .language mismatch signal (metric) is zero
to one hundred
percent, where one hundred percent would have indicated that all campaign
participation actions
in this cluster/campaign are Tweetedrm (or otherwise posted) in a language
different from their
accounts' default profile language.
104501 How the language mismatch signal (metric)Is computed For each TweetTm
Or other
posting) with the campaign-related hashtag, the language mismatch signal
(metric) may identify
the language of the Tweet (Or other posting) and the language profile setting
in the Twitterrm
API or the API of another social media platform, in embodiments; the language
mismatch signal
(metric) may also aggregate the Tweets' l (or other postings) by the cluster
of the author of the
Tweet"' (or other posting) in a campaign map. By way of this example, the % of
Theetsm (or
other postings) for each cluster whose tweet language did not match the poster
language of the
1'weetTM (or other posting) may be reported.
104511 Detailed embodiments of the present disclosure are disclosed herein;
however, it is to be
understood that the various disclosed embodiments are merely exemplary. of the
disclosure, which
may be embodied in various forms. Therefore, specific structural and
functional details disclosed
herein are hot to be interpreted as limiting; but merely as a basis for the
claims and as a
representative basis for teaching one skilled in the art to variously ernploy
the present disclosure
in virtually any appropriately detailed structure.
10452] The terms "a" or "an,'' as used herein, are defined as one or more than
one. The term
"another," as used herein, is defined as at least a second or more. The terms
"including" and/or
"having." as used herein, are defined as comprising (i.e., open transition).
92
[0453] While only a few embodiments of the present disclosure have been shown
and described, it will be
obvious to those skilled in the art that many changes and modifications may be
made thereunto without
departing from the spirit and scope of the present disclosure as described in
the following claims.
[0454] The methods and systems described herein may be deployed in part or in
whole through a machine
that executes computer software, program codes, and/or instructions on a
processor. The present disclosure
may be implemented as a method on the machine, as a system or apparatus as
part of or in relation to the
machine, or as a computer program product embodied in a computer readable
medium executing on one or
more of the machines. In embodiments, the processor may be part of a server,
cloud server, client, network
infrastructure, mobile computing platform, stationary computing platform, or
other computing platform. A
processor may be any kind of computational or processing device capable of
executing program
instructions, codes, binary instructions, and the like. The processor may he
or may include a signal
processor, digital processor, embedded processor, microprocessor, or any
variant such as a co-processor
(math co-processor, graphic co-processor, communication co-processor and the
like) and the like that may
directly or indirectly facilitate execution of program code or program
instructions stored thereon. In
addition, the processor may enable execution of multiple programs, threads,
and codes. The threads may be
executed simultaneously to enhance the performance of the processor and to
facilitate simultaneous
operations of the application. By way of implementation, methods, program
codes, program instructions
and the like described herein may be implemented in one or more thread. The
thread may spawn other
threads that may have assigned priorities associated with them; the processor
may execute these threads
based on priority or any other order based on instructions provided in the
program code. The processor, or
any machine utilizing one, may include non-transitory memory that stores
methods, codes, instructions, and
programs as described herein and elsewhere. The processor may access a non-
transitory storage medium,
through an interface that may store methods, codes, and instructions as
described herein and elsewhere. The
storage medium associated with the processor for storing methods, programs,
codes, program instructions
or other type of instructions capable of being executed by the computing or
processing device may include
but may not be limited to one or more of a CD-ROM, DVD, memory, hard disk,
flash drive, RAM, ROM,
cache, and the like.
[0455] A processor may include one or more cores mat may enhance speed and
performance of a
multiprocessor. In embodiments, the process may be a dual core processor, quad
core processors, other
chip-level multiprocessor and the like that combine two or more independent
cores (called
93
CA 3068264 2022-08-03
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
a die).
104561 The methods and systems described herein may be deployed in part or in
*hole through a
machine that executes computer software on. a. server, client, flrewall,
gateway, hub, router, or
other such computer and/or networking hardware. The software program may be
associated with
a server that may include a file server, print. server, domain server,
Internet server, intwet server,
cloud server, and other variants such as secondary server, host server,
distributed server, .and the
like. The server may include one or more of memories, processors, computer
readable media,
storage media, ports (Physical and virtual), communication devices, and
interfaces capable of
accessing other servers, divas, machines,-:and devices through a wired or a
wireless medium, and
the like. The methods, programs, tir.:cOdit as described herein and elsewhere
May be executed by
the server. In addition, other devices required for execution of methods as.
described in this
application may be conSidered as a part of the infrastructure associated with
the server.
104571 The server may provide an interface to other devices including, without
limitation, clients,
other servers, printers, database servers, print servers, file servers.,
communication servers,
distributed servers, social networks, and the like. Additionally, this
coupling -and/or connection
May facilitate remote:execution of program across the network. The networking
of some or all of
these devices may facilitate_ parallel processing of a program or method at
one or more location
without deviating from the scope of the disclosure. In addition, any of the
devices attached to the
server through an interface may include at least one storage medium capable of
storing. methods,
:programs, code and/or instructions. A central repository may provide program
instructions to be
executed on different devices. In this implementation, the remote repository
may act as a storage
medium for program code, instructions, and programs.
104581 The software program may be associated with a client that may include
.a file-client,: print
Client domain client, Internet client, intranet client and other variants such
as secondary client,
boat-Client, distributed .client, and the like,. The client may include one or
more of memories,
processors, computer readable media.õ Storage:media, ports (physical and
virtual), communication
devices, and interfaces capable of accessing other clients, servers, machines,
and devices through
-a wired or a. wireless medium, and the like. The methods, programs, or codes
as described herein
and elsewhere may be executed by the client In addition, other devices
required for execution of
methods as described. in this application .may be 'considered as a part: of
the infrastructure
associated with the client.
104591 Theelient may provide an interface to other devices including, without
limitation, servers,
other clients,, printers, database servers, print servers, file, servers,
communication servers,
distributed servers, and the like. Additionally, this coupling and/or
connection may facilitate
remote execution of program across The network. The networking of some or all
of these devices
94
CA 03068264 2019-12-20
WO 2018/237098 PCT/US2018/038639
may; fatilitate. parallel processing of a program or method at one or more
location without
deviating from the scope of the disclosure. in addition, any of the -devices
attached to the client
through an interface may include .at- least one storage medium capable. of
storing methods,
programs, -applications,, code and/or instructions, A central repository may
provide program
instructions to be executed on different devices. in this implementation, the
remote repository
may act as a storage medium for program code, instruction* and programs.
104601 The methods and systems described herein may be deployed in part or in
whole through
network. infrastructures. The network infrastructure may include elements such
as computing
devices, servers, routers, hubs, firewallsõ clients, personal- computers,
communication -devices,
.routing devices and other active and passive devices,. modules and/or
components as known in the
art. The computing and/or non-computing device(s) associated with the network
infrastructure
may include, apart from other components, a storage medium such as flash
memory, buffer, stack,
RAM, ROM, and the like. The processes, methods, program codes, instructions
described herein
and elsewhere_ may be executed by one or more of the network infrastructnral
elements. The
methods and systems described herein may be adapted for use with any kind of
private,
community, or hybrid cloud computing network or cloud computing environment,
including those
-which involve features-of software. as a service (SaaS), platform as a
service (Pa.aS), and/or
infrastructure ass service(laaS).
104611 The methods, program codes, and instructions described herein and
elsewhere may be
implemented on a cellular network having multiple cells. The. cellular network
may either be
frequency division multiple. access (FL)MA) network or code division multiple
access (COMA)
network. The cellular network may include .mobile devices, cell sites, base
stations, repeaters.
antennas, tower* and the like. The cell network. may be a GSM. GPRS, 3G, ENDO,
mesh, or
other networks types.
I04421 The methods, program codes, and instructions- described herein and
elsewhere may be
imPlemerded on or through mobiledevices. The mobile devices may include
navigation devices,
cell phones, mobile phones, mobile personal digital. assistants, laptops,
palmtops, netbooks,
pagers, electronic books readers, music players and the like. These devices
may include, apart
from other components, a storage medium such as a flash memory, buffer, RAM,
ROM and one
or Mort computing devices. The computing_ devices associated with mobile
devices may he
enabled to execute program codes, methods, and instructions stored thereon.
Alternatively, the
mobile devices may be configured to execute instructions in collaboration with
other devices. The
mobile devices may communicate with base stations interfaced with servers, and
configured to
execute program codes. The mobile devices may communicate-on a peer-to-peer
network, mesh
network, or other communications network. The program code may be stored on
the storage
DEMANDE OU BREVET VOLUMINEUX
LA PRESENTE PARTIE DE CETTE DEMANDE OU CE BREVET COMPREND
PLUS D'UN TOME.
CECI EST LE TOME 1 DE 2
CONTENANT LES PAGES 1 A 95
NOTE : Pour les tomes additionels, veuillez contacter le Bureau canadien des
brevets
JUMBO APPLICATIONS/PATENTS
THIS SECTION OF THE APPLICATION/PATENT CONTAINS MORE THAN ONE
VOLUME
THIS IS VOLUME 1 OF 2
CONTAINING PAGES 1 TO 95
NOTE: For additional volumes, please contact the Canadian Patent Office
NOM DU FICHIER / FILE NAME:
NOTE POUR LE TOME / VOLUME NOTE: