Note: Descriptions are shown in the official language in which they were submitted.
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
TITLE
METHOD OF SOCIAL NETWORK GENERATION
CROSS-REFERENCE TO RELATED APPLICATIONS
Under 35 U.S.C. ~ 119(e), this application claims the benefit of U.S.
Provisional
Application Serial No. 60/134,018 filed May 12, 1999. A portion of the
disclosure of
this patent document contains material that is subject to copyright or
trademark
protection. The copyright or trademark owner has no objection to the facsimile
reproduction by anyone of the patent document or the patent disclosure, as it
appears in the Patent and Trademark Office patent file or records, but
otherwise
reserves all copyright or trademark rights whatsoever.
FIELD OF THE INVENTION
The present invention relates, in general, to database processing, and, more
particularly, to systems and methods for organizing and analyzing customer
records.
BACKGROUND OF THE INVENTION
The current focus within the financial services industry (banking, insurance,
brokerage) and other service industries (e.g., telecommunications,
hospitality) is the
concept of customer relationship management. This focus highlights the
customer
as the nucleus of the relationship and suggests that all interactions between
the firm
and the client need to be managed at a customer level, rather than at a
product
1
CA 02374120 2001-11-09
WO 00/68860 PCT/IJS00/13132
level. Prior to the 1970s the financial services industry took a more account-
oriented approach to managing their customers - that is, they viewed the
customer
as merely an extension of the account (or policy) rather than as the focal
point of the
business relationship. "Knowing the customer" meant simply that you could
identify
who was related to a given account as the primary owner in order to be able to
address marketing and statement literature. As the ability to group accounts
together into "households" (see definitions) became technologically feasible,
awareness began to emerge of the need to understand more about the market
potential for cross-selling to existing customers while at the same time
creating a
more comprehensive profile of each customer as an individual entity with which
the
firm conducts business. This is important for the purposes of risk management,
retention management, pricing, channel balancing, sales management, and
general
customer servicing. While this abstract talks about the financial services
industry,
the problem has analogies for all types of firms for their business-to-
business
relationships and their firm-to-retail customer relationships. In addition
there is an
analog in the pharmaceutical industry and others to which this technology may
be
applied.
Companies and organizations find it critical to their success to understand
the value
of their customers ("Customers" in this disclosure shall also include
prospective
customers), inclusive of all opportunities and risks, and wish to manage the
relationships they have with customers based upon this understanding. They do
this
through analyzing a variety of data pertaining to customer accounts, including
products or services purchased, payment histories, and sales or performance
histories. This information is held in their own data repositories. Further,
2
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
organizations may link their databases to other outside data sources to enrich
their
information on the customers. The resultant information will be useful in the
course
of managing the company or organization's relationship with their customers
(and
prospects). This data can be used in a variety of ways to value and manage
customer relationships. Using this data, companies can customize how the
customer is handled based on their assessment of the customer's value (which
may
include influence over other customers) to the company or organization and any
associated credit risk. The determination of the value of the customer is used
to
make decisions about a number of issues including, but not limited to, the
pricing of
services provided or to be offered, the types of products and services to be
offered,
the level of service offered, and the amount of credit granted.
To date, identifying a customer's overall relationship and viewing data
relating to the
customer has been achieved through the development of databases designed to
accept extracts of data from the variety of accounting/information systems a
company or organization uses to administer accounts (so-called "accounting
application" systems). For example, in a multi-division company, central
repositories
may be used to determine how much business each division does with a customer
and what the customer is worth to the holding company as a whole. These
databases can also accept data from outside sources. To date the process used
to
bring this data together to create this unified view has relied primarily on
matching
names and addresses.
By using these methods of creating databases, companies and organizations can
"see" all the accounts (and related data) for each customer at a specific
location
(defined as a Marketing CustomerT"'). However it is also important to
understand
3
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
relationships and value them at the level of the economic buying unit.
Consequently
current technology uses variations on the name and address matching techniques
used to identify and link a customer's relationships to each other and outside
data to
group customers living at the same address; this process is called
"householding."
The business-to-business analogue groups companies together that have owners
in
common. This is called a "corporate household."
Householding is important, because for some products if there is more than one
individual living at a specific mailing address, a product may be the type
that all the
individuals may not buy. For example, no more than one first mortgage can be
outstanding on a single property, and everyone living in a single home or
apartment
will use the same washing machine, dishwasher, long distance service on a
phone
line and so forth. It thus makes no sense to make an offer for such a product
or
service to multiple people living together at an address or to offer a second
person
living at the same address another of such a product when someone else living
there
is already using it. In other cases it does make sense to make offers of
additional
products or services or other products or services a company or organization
might
offer to or for the use of others living at the same address. For example if
there are
two adults living at an address sharing a checking account, but only one ATM
card
and one credit card not held jointly, the person or persons not possessing
these
products would be a good candidate to be approached about acquiring these
products.
Concurrent to this evolving marketing view of customers was the development of
customer information files (CIFs) which were constructed in order to identify
all of the
customers directly linked to an account or policy or product application
system
4
CA 02374120 2001-11-09
WO 00/68860 PCT/iJS00/13132
purchased by the customer. This would serve to couple the customer to all
products
they had purchased, allowing the firm to conduct more effective analysis on
customer behavior. This identification of single customers (either commercial
or
retail) also facilitated more intelligent Householding, as it would become
possible to
group customers (rather than simply accounts) into meaningful economic
decision
making units. Absent from these new approaches, however, was a means of
constructing linkages between customers that were less obvious than the fairly
simple association of surname and address (the definition of a marketing
household). As the structure of society changed (increased divorces, greater
numbers of the presence of children from a previous marriage, higher numbers
of
women retaining their maiden name, etc.), it became necessary to be able to
build
relationships between two individuals who may not share the same surname but
who
may still live at the same address. This kind of "contemporary or extended
Householding" required a modification in the manner in which customer linkages
could be deciphered, but remained dependent upon an address similarity to
actually
infer a link. While this was a major step forward, it continues to provide
limitations to
truly understanding the nature of a clients' overall network of relationships.
At present, the values assigned to customers and the relationship management
strategies pursued are arrived at using currently available systems whereby
the
customer and all others residing at a specific address are defined as a
"household"
for analytical purposes. A customer can only have one address and a unique
last
name under currently used systems (defined in this application as a Marketing
HouseholdT"~). There are commercially available computer routines that use a
series
of commands to discern which customer or customers reside at a specific
address
5
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
(even if the data fields are misspelled). All the company or organization's
databases,
often referred to as "accounting applications" containing information on
product or
service sales, balances, losses, or any other relevant information, are then
sifted
through. Information useful for marketing and relationship management on all
persons residing at a specific address is extracted from the accounting
applications
and is used in the creation of a database commonly referred to as a "Customer
Information File," or "CIF."
The CIF contains information about all the accounts or products held or owned
by
people residing at the specific address and other data the company or
organization
deems relevant to managing the relationship from its own records. It may be
enriched with data from outside the company or organization's files that is
helpful to
determining the customer's value or potential value and how the relationship
should
be managed. The value of the relationship in such a scheme can be looked at
from
an entire household basis or on a customer-by-customer basis within the
household
(but also as defined by a specific mailing address for each customer).
By virtue of how the data is organized in these schemes and the inherent
restrictions
of defining a relationship by attaching it to a specific address, this
approach has
several restrictions. For example, the current state-of-the-art software
assigns the
value of a specific account to the person whose name first appears on the
accounting application from which the data is extracted in cases where more
than
one name may appear on the account record. This person may or may not be the
decision-maker with regard to the account in question or any other accounts
that the
household in question may consider buying. Additionally this person may be
considered to be two or more people. For example, a single customer who uses
or
s
CA 02374120 2001-11-09
WO 00/68860 PCT/IJS00/13132
has used different names at different times (for example married, maiden,
hyphenated, and so forth) would be considered different customers from a
person at
the same address. Spanish surnames may have one or two surnames and are
particularly difficult.
More shortcomings are a result of the inherent limitations of tying the
organization of
the data to a specific address. Most notably, accounts controlled or
influenced by a
customer residing at one address but listed on accounting applications at
another
address are excluded. Several examples of account and/or influence
relationships
that may be missed follow:
A. Business accounts. A person residing at one specific address may own or
control a business account. The business, with the exception of home-based
businesses, will have its own specific address and will be listed in the
company
name. The company name is almost always different than the individual's name.
B. Accounts over which the person has influence, but where that person is
listed on
one account at an address but is not living at the same physical address.
Examples of this sort of situation include, but are not limited to:
i. an elderly couple where the wife is living at home and the husband resides
at a nursing home;
ii. a child of divorced parents with two possible addresses and two possible
co-owners;
iii. cosignatory or trustee addresses;
iv. children at college;
7
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
v. accounts held jointly with parents or others residing elsewhere where the
other party appears first on the accounting application and whose mailing
address information is held on file;
vi. accounts for which someone is a cosigner or guarantor for others;
accounts administered for a person by a bank trust department, lawyer or
accountant, and so forth.
This problem is magnified where files hold records for owner, administrator
and
mailing name/addresses (common in Britain and Ireland).
C. In other cases influence may be indirect, but can be just as real. For
example
co-workers at a work place can influence one another, or a business owner can
agree to allow his employees to be solicited for a product or service. A great
deal
of products and services are sold via personal recommendations. Current
systems are completely incapable of identifying relationships where a person
has
the potential to influence another person other than if they physically share
an
address and there does not exist any measure of the value of the customer in
terms of customers to whom he is directly connected.
D. Accounts held by the same person, but at another address such as where the
customer has multiple addresses due to owning multiple properties. The person
may also use a Post Office Box or a work address in addition to a "home"
street
address in some circumstances. For example, a person or family may own three
homes (primary residence, lake getaway cabin, and warm weather home they
frequent during the winter months). This person or family may have three
mortgages all with the same company. These relationships are missed by
s
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
current systems. A more expansive and accurate definition of household lowers
the customer/household count and raises the average number of products sold
per customer. The process and technology of the invention allows for such a
definition to be used and by correcting the problem inherent in current
systems
and provides more accurate counts of customers/households. The results give
more accurate measures of product cross sell.
Similar dynamics occur where businesses are concerned. The current state of
the
art uses the same address and name-based matching and grouping techniques for
businesses as are used for consumers. The business segment is prone to the
same
issues as the consumer segment that cause a complete picture of the value of a
customer not to be able to be developed using current techniques. Companies
and
their subsidiaries may have multiple names and addresses that current
methodology
is unable to recognize and group for analytical purposes. The influence over
relationships is thus unable to be tracked.
The next step in the evolution of customer relationship management is to
understand
the full scope of a client's associations to other clients, whether these
relationships
be of an individual-to-individual, individual-to-business, business-to-
business or
business to individual nature. The schematic developed from this linking is
referred
to as a "social network" or "Customer NetworkT""". The primary importance in
developing social networks is to be able to comprehend the "Influence
ValueT"'"
which a client reflects. This measurement is derived by aggregating some value
associated with each client (balance, profit contribution, etc.) and assigning
that
aggregate amount to each customer in the relationship as a measure of the
amount
of business (or profit, etc.) over which they may exert influence. It is
intended to
s
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
provide insight into the sphere of influence that a customer has over other
customers
(or prospects) with which a firm may do or wish to do business. This type of
information is useful in a variety of relationship management areas that
include but
are not limited to assessing risk, in determining propensity to buy or in
managing
customer retention.
Building these network links is facilitated by the analysis of a set of
characteristics
that may be shared by two or more customers. Examples include joint account,
phone numbers, tax id numbers, employer or any other distinct element of data
captured in a customer record, etc.
Thus the challenge is to determine the appropriate metrics with which to
monitor
customers within their given social network. Further, decisions must be made
about
the use of the data derived by building not only the network itself, but also
by the
aggregation and assignment of such metrics described above.
In the 1960s, organizations viewed their business through the eyes of their
product
managers and did not examine customer relationships. It was in this decade
that
customer files were in their infancy. In the 1970s, enterprises began to
attribute
metrics to customers, such as sales, deposits, loans, and profits. In the
1980s, there
was a move to defining households. Businesses recognized that many purchases
were made for the household. A family had one mortgage, one stove, etc. Trying
to
reach every person in the household with direct marketing was overkill. A
family with
five members was still going to have only one mortgage. A term was developed
called Economic Buying Unit (EBUT~"). A household was defined as a unique last
name occurring at a unique address. Measures were then developed for cross
sell
~o
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
to show the number of products purchased by a household and the value of these
products.
Households and customers as defined above shall be categorized as marketing
customers and households. These definitions have several shortcomings. In
research on social networks, these shortcomings came to light. There were many
people living at the same address with different last names that shared joint
accounts. Customers had multiple addresses. An effort was undertaken to
determine what was causing customers and households to be categorized
incorrectly.
For example, married women began to have business relationships under both
their
married and maiden names. Divorced women remarried, and their children kept
the
father's name yet they lived in the new husband's domicile. These people are
really
in the household and are really the same customer. This led to constructing
networks of people who lived at the same address and shared an account. This
led
to the term Real CustomerT"' and Real HouseholdT~". If they shared an account
and
lived at the same address and did not share a social security number, then
they were
considered to be in the same household. If a customer is female, has the same
social security number and last name - or has a different last name, the woman
is
assumed to be using her maiden name and is the same person.
Carrying the concept further allows one to look over customer data sets for
people
with multiple addresses who share all the name fields and tax identification
numbers
with other customers. If so, they are the same customers and in reality there
is only
one "true customer". (If they are female, the last name is permitted to be
different.)
11
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
Analyzing data this way reduces the number of customers and households and
gives
a much truer picture. With over counting of customer and households,
institutions
are underestimating the cross sell ratio of products per customer and
household.
For the customer and household over counting, this could vary in actual
situations by
between two and fourteen percent, depending on the geographic area and the
data
quality. This affects the cross-sell ratios greatly and can lead to incorrect
decisions
in marketing, de-marketing and pricing.
To solve these problems, the invention claimed herein creates customer social
networks and the calculation of "influence value" where the above shortcomings
are
avoided by setting up a database that has a clustering process and indexing
methodology that focuses on data available at the account level other than
name
and address. Further, a value is developed for each customer that measures
individual value and the value of related customers. From these techniques
large
numbers of relationships may be developed and large networks of people and
organizations can be developed. These networks can then be used to configure
sales and marketing channels.
Organizations often attempt to measure the percentage of their customers that
account for 80 percent of their profits. One rule of thumb is that 20 percent
of the
customers accounted for 80 percent of the profits. Through de-marketing,
service
fees and other techniques, as few as 8 percent of the customers can account
for 80
percent of the profits, and it is possible that in reality, half of this 8
percent could
influence 80 percent of the profits. The technique of the present invention
has the
capability for identifying this group of individuals and thus giving companies
~2
CA 02374120 2001-11-09
WO 00168860 PCT/US00/13132
tremendous understanding of their customers and how sales channels may be
configured to give them optimal service. With existing methodologies, customer
data is examined at a point in time. This data may be compared then to
customer
data at an earlier point in time. For example, assume that there are two
million
customers and there were a hundred thousand less in the previous time period.
With time-dated referential history, it is possible to generate databases that
have the
same customer and household composition across time periods. This allows for
changes to be more exactly calibrated and for buying behavior to be compared
for
the same customer and household composition. The change in customers between
time periods is really the net between customers lost and customers gained.
The
same is true with sales figures across time periods. Sales may be increasing
in the
aggregate, but this is the net against those customers with declining sales,
lost
customers, new customers, and customers buying more. This technology allows
the
customer database composition of customers and households to be consistent.
SUMMARY OF THE INVENTION
The present invention relates, in general, to a relational database, and, more
particularly, to systems and methods for taking accounting application data
and
organizing it into information by customer, household, and associated
customers
(networks) and analyzing customer records and associated buying patterns. The
method of analysis depends on "influence", which is a defined term developed
herein. The analytical technique described herein develops a measure for
customers and households which measures the influence an individual customer
may have on purchasing within a network of connected customers.
13
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
The invention uses a new, unique and powerful methodology to identify all the
relationships a customer has with a company or organization and the
relationships a
customer has with other customers or prospective customers of the company or
organization. It also allows for links that exist within existing data files
to be followed.
This means indirect associations can be considered and acted upon. The
invention
gives companies a more accurate view of their customers and allows them to
ascribe more pertinent values to them and to create and pursue better customer
and
credit management strategies. Furthermore it does so in a manner that is
efficient
from a data processing perspective, thus minimizing the resources required and
costs of producing the results of the invention.
The process uses social network theory and non-address-based matching
techniques to allow two substantial improvements in defining the value of a
customer
to be derived. The first is the result of the process employed whereby the
"true"
value of the customer is arrived at through the inclusion of accounts over
which the
customer has control at any and all addresses including businesses.
The second is called Influence ValueT"'. The same matching techniques used to
discern the true value of a customer are used to find links a customer or
household
has to others (the person's, company or organization's or household's so-
called
social network). This network is called the Customer NetworkT"' or Client
NetworkT""
since it is not necessarily an individual as the word "social" would imply
Using this
technique the analysis of the value of a customer or a traditional household
can be
expanded to include the size and value of the person's, company or
organization's or
household's network one or more links (or levels or direct links) away. For
example,
14
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
an employee of a company would be one link away (defined as influence of the
first
degree and also indirect or affinity influence). All the people to whom that
employee
had links, such as his or her spouse and children, would be two links away
(influence
of the second degree) from the employer, unless the spouse or children also
worked
for the same employer.
The result is that companies can develop more accurate views of the value of
customers. A new paradigm of customer or household value is created that
allows
for different and more effective customer management strategies to be
employed.
By way of example, consider a comparison of a view of members of a family
using
the current state of the art and the invention. This family, described briefly
previously
as an example of the type of important information missed by current systems,
has
three addresses - a summer weekend getaway home, a principal residence, and a
residence in Florida that is the customer's principal residence for a four-
month time
period. The family also owns a small business for which the wife is the
bookkeeper,
and the customer's wife has a large trust account administered through a law
firm.
The wife also has signature authority on one of her wealthy elderly mother's
accounts.
Currently the wife could be viewed as up to six different "customers" in six
different
"households." The wife could be a "different" customer at each of the three
addresses (principal residence, weekend getaway, and the Florida home) the
business address of the family business, her mother's address, and the law
firm's
address. The number of times she is considered a customer would, using present
technology, furthermore be determined to a large degree by chance based on
which
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
accounts she happened to be listed as the first name on the accounting
application.
Standard analytical techniques today attribute all customer value to the
person/organization on line one of the account.
On their own, none of the current "customer" or "household" or "marketing"
systems
would identify the wife in this example as deserving of special treatment. In
fact,
some or all of what appear to be separate customers or households, using
currently
available tools, might be subject to strategies seeking to make "unprofitable"
customers profitable or to encourage them to close their accounts through the
imposition of high fees (de-marketing) on the account or accounts identified
through
the limited view provided by the current state of the art. The right strategy
would be
to provide this customer special treatment and perhaps even reduce some fees
based upon the customer's value. In a test project, a signatory on a privately
held
business account with two million dollars in certificate of deposits, had his
fees
increased because his personal checking account did not maintain the proper
balance.
None of the value calculations that a company could execute using current
technology would take note of the large trust account or the wife's influence
over
other valuable accounts. Imagine the damage that could be done were such a
negative action to be taken on just one of the several accounts. The customer
would
likely be displeased by the negative action and might close the account in
question
and a series of other accounts. Furthermore it is very possible that the
customer will
mention their displeasure to those in their network (including the law firm
which may
have recommended the bank to the customer in the first place). These people or
organizations may hold accounts or may be considering opening accounts at the
16
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
company or organization or, as in the case of the law firm, be in a position
to
recommend the company to others. (People who are unhappy may tell three people
on average, while a happy person on average tells less than one person on
average.) The best case is that the customer will take no action, but will
have
developed a negative perception of the company that leaves them open to
considering offers from the company's competitors.
On the other hand, imagine the good that would be done if the correct strategy
that
pampers this customer were employed. The customer would be less likely to
accept
offers from competitors or to switch their accounts to the competitor and
would be
more likely to consolidate even more accounts or purchase more products with
the
company because of the preferential treatment that is being provided.
In the same scenario as above, the invention would save money for the company
or
organization using it by allowing it to treat this customer as one customer
instead of
as multiple customers. Rather than sending solicitations for a credit card to
multiple
addresses simultaneously, solicitations would be sent to one address at a time-
only.
In test usage the invention has shown that the current state of the art
overestimates
the number of customers and households by 7-9%.
To obtain the results of the invention the process followed is that starting
with
accounting application names and addresses, each line that has an individual
or
company name is assigned a unique identifying number or customer number. A
customer record is generated for each customer with the address of the source
record. These records are then matched for duplicates, and pointers are set in
a
customer to accounts cross-reference list. For records that have the same
y~
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
customer last name and address, a Marketing HouseholdT"" is created with a
list of
customers in the household.
For each customer, a list is also created that contains customer information
for
identifying elements that are unique (meaning that only one such valid record
can
exist even though in some cases multiple people may be associated with the
element). These elements include telephone numbers, email addresses, account
numbers, employer identification number in the case of companies or
organization,
and social security number. In addition, customers who share accounts have
account numbers in common. Also on the list by customer is the employer
identification number, telephone number and email address of the customer. If
the
customer is a company, the customer is linked to a file such as that available
from
Dun & Bradstreet. From this, a list of customer to headquarters,
establishment,
parent, and ultimate Dun & Bradstreet number is built.
Whether the customer is an individual or a company or organization the file
created
is then sorted by data point, then customer number. The file is then traversed
by
data point. Records that have equal data points produce a new work record
consisting of pairs of serial numbers; these serial numbers are those of the
customers sharing a common data point such as employer identification number
or
telephone number.
This record with the paired serial numbers is then passed through a process
that
builds the chain of all of the records that share common serial numbers. The
lowest
serial number becomes the unique cluster identifier. Since the original serial
number pairs were developed based on a data point sort, the same serial number
18
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
may be present in various points in the file. This requires re-sorting and
iterative
runs of the process to resolve all of the links of the chain and the
assignment of the
final cluster number. The iterations continue until no more links are
resolved. The
output of this step is a record with the serial number and the cluster number
(social
network).
This file is then joined back with the file containing the serial numbers
pairs. This
establishes the relationships or linkages in the network. Joining back with
the
database's customer identifier produces the links (called edges in some
literature).
The database also houses "No LinkT"'" pointers. The purpose of these pointers
is to
prevent linkages which occur in an automated fashion but which should not come
together. For example, a child support account for the ex-wife should not be
linked
into the same household as the present wife, yet with current technology, it
would be.
"No links" are hand coded.
Further, the approach keeps track of aliases of names. For example, matching
technology can find nicknames (Bob versus Robert) and maiden names of
customers. The invention auto generates alias records to speed future location
and
inquiry, regardless of the name used when customer records are combined.
From a data processing perspective, once the master database is built, a
process
whereby the database is updated to reflect changes on an incremental basis is
used.
Present technology requires extraction of all names and addresses. Lists of
fragments of customer records are indexed back to the source records.
Fragments
are developed from new customer records. The intersection of hits on fragments
in
common from the new record with records on the database points to a target
list of
19
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
duplicates, which then may be evaluated to determine the most likely
duplicate. The
invention posts name and address maintenance back to other applications based
upon parameters and incorporates a methodology to keep multiple customer and
application systems synchronous (for example, a data warehouse and an
operational
data store). This is defined as the distribution process or DistributorT"~.
The
DistributorT"' process avoids the need for a total regeneration of data sets
when
refreshing the system with new data. This approach automatically keeps the
warehouse, the application systems, and the central operational database
synchronous in terms of customer name and address information.
The invention through the lists of links it creates and the way in which it
organizes
the data provides flexibility to analyze and use the data in a large variety
of manners.
This is done by including or excluding certain types of linkages and grouping
or
ungrouping people and households and consumer and corporate or organization
accounts. Unlike current systems, the invention's system is not constrained by
a
reliance on the use of a single customer at a physical address (or multiple
addresses) to create and organize associations.
Where the old view of the world is apropos and it does make sense to market to
a
single physical address (for example, in the marketing of a mortgage that must
be
secured by a specific property), the new invention can identify the most
logical
person to whom the offer should be addressed based upon household attributes.
The current state of the art assigns the recipient based on which person
residing at
the address' name happens to appear first on the record. The system of the
invention would allow the user to use the data to decide which household
member is
the most likely to respond to an offer. With this invention, the user
specifies with
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
which unique names at which unique addresses to correspond. A key with this
invention is that the identification and creation of the links and the
database
organization allows users to target customer management and marketing programs
based upon the appropriate definition of the customer or household the
activity
envisions.
The invention can segregate influence on whether the relationship is
individual-to-
individual, individual-to-organization, organization-to-organization, or
organization-to-
individual. This allows users of the invention to target promotions or
customer
management strategies using these attributes and associations that were
heretofore
not systematically identifiable. For example a user could use the invention to
fashion
a program to market a business-banking product based on the network
affiliations
identified by the invention.
Another way to manipulate the data for business-to-business applications is by
whether the relationship is "direct" (i.e., the person has a direct link to an
account; for
example, they have signature authority on the account) or "affinity" (for
example, a
person who works at the business and is potentially influenced as a result of
their
association with the business). Affinity RelationshipsT"' occur when the
organization
is on line 2 or greater of the accounting application from which the data is
culled. In
the case of a direct relationship, the invention's user would be able to
target Direct
RelationshipsT"~ to be sold additional products and services the business
might use,
for example, a business loan. In the affinity case, a bank might offer the
people with
an Affinity Relationship no minimum balance free checking with direct deposit
of their
paycheck into an account with the bank.
21
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
In summary the invention allows users to:
avoid undervaluing a customer when information such as account balances at a
single physical location alone can be misleading or no value is given the
customer if their name happens to not be on line one of the account.
. establish a customer valuation process that provides compelling evidence
based .
upon all the facts available and allows the user to make business decisions
based upon a true picture of customer value that also takes into account the
customer's network associations and Influence ValueT"'
understand the full extent of customer-to-customer relationships and
automatically establish customer-to-customer links in order to more accurately
represent metrics based on such linkages (i.e., cross-sell ratios, product to
customer ratios, product to household ratios, or household credit usage).
link one individual to others to form a network of relationships (and
marketing
opportunities) that may be otherwise unknown.
. identify individuals with influence over other people or businesses and use
this
information in sales and marketing programs.
apply to large customer bases and in an automated fashion one-to-one customer
relationship management. The invention allows users to access and use the type
of knowledge a small town bank manager would have of his or her customers -
their associations with each other, and their level of influence as measured
by the
value of business related parties do with the bank through their extended
families
and/or community associations.
22
CA 02374120 2001-11-09
WO 00/68860 PCT/L1S00/13132
This invention may be extended from a business to consumer or business to
business sales, marketing and customer management to any problem where
networks of relationships might exist. One example of such an extension is in
the
area of pharmacology.
Other advantages of the present invention will become apparent by a perusal of
the
following detailed description of presently preferred embodiments of the
invention
taken in connection with the drawings.
BRIEF DESCRIPTION OF THE DETAILED DRAWINGS
Figure 1 shows a compound that occurs in two drugs
Figure 2 shows the family relationship being collapsed to diseases.
Figure 3 shows an example of a social network.
Figure 4 is a process flow chart showing the steps in the system of the
invention.
Figures 5 and 6 are examples of a visual display of a Customer NetworkT"~ as
created by the invention and viewed using the software described in the
preferred
embodiment.
Figure 7 illustrates the Database Creation procedure of the invention
Figure 8 illustrates the Identify Relationship Hierarchies procedure of the
invention
Figure 9 illustrates the Ongoing Record Maintenance and Post and Distribute
procedures of the invention
Figure 10 illustrates the Attribute Account Information to Marketing Customer
and
Familial Collapse (Part I) procedures of the invention
Figure 11 illustrates the Familial Collapse (Part II) procedure of the
invention
23
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
Figure 12 illustrates the Familial Collapse (Continued) procedure of the
invention
Figure 13 illustrates the Influence Value calculation procedure of the
invention
Figure 14 illustrates the Customer Reporting and Analysis procedure of the
invention
Figures 15 and 16 show a diagram of the communication patterns in a company.
Appendix A contains a list of definitions of terms used in the describing the
invention
as set forth herein.
Appendix B provides an example of elements of a programming architecture that
can
be used to implement an embodiment of the invention
24
CA 02374120 2001-11-09
WO 00/68860 PCT/~JS00/13132
DETAILED DESCRIPTION OF THE PRESENTLY PREFERRED EMBODIMENTS
Figure 3 depicts a network of individuals and organizations that have
associations or
"relationships" with each other. The invention is comprised of a method for
processing data using computer readable code (software) configured to cause a
computer to process data such that the lists of relationship links are
generated.
These lists in the preferred embodiment are then manipulated using a program
exhibiting functionality similar to that of MatRes (software developed by
Innovative
Systems, Inc.) and which can be viewed graphically using software exhibiting
functionality similar to that developed by Tom Sawyer known as Graph Layout
Toolkit 2.4. Figure 5 is an example of a visual display of a Customer
NetworkT"~ as
created by the invention and as viewed by the software. The data can also be
exported to any of a variety of software packages for analysis, use, and
viewing of a
non-visual nature.
The invention consists of the process that is used to generate the lists of
the
relationships that are used to create a consolidated view of each customer
across
multiple attributes important to (and selected by) the user and to link
customers to
other customers and the resultant ability to analyze and measure the Influence
ValueT"' of a customer or household. As previously described, the lists can be
used
to develop visual diagrams of relationships that can be used for various user
activities including but not limited to sales, marketing, customer service,
credit or risk
evaluation or customer retention. They can also be used to select records for
particular handling. The lists and visual diagrams can be organized by a
variety of
definitions of customer and household that are more or less expansive to
create the
most appropriate database view against which to perform the aforementioned
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
activities. The lists can also be used to generate a measure of Influence
ValueT"",
and this measure can be calculated one or more links away from the target
customer
or household and again can be based upon whatever definition (narrow or
expansive) of customer or household the user chooses. One link away is defined
as
the first degree of influence. For purposes of this description, a degree is
defined as
the number of links away from the customer in question.
Further, the networks can be collapsed by household or by degree and influence
computed, then, by household or degree (Familial CollapseT"").
Figure 4 is a process flow chart showing the procedures used in creating the
system
of the invention. These procedures are further broken down into steps that are
described below with reference to Figures 7 through 14.
Procedure 1 -- Database Creation
This procedure defines customer relationships on a Contact record and loads
the
resultant output to a relational database (see Figure 7).
Step 1 A Extract (Account & Contact Data)
The process is started by loading all account and prospect information from
the
account prospect systems that are going to be used in the marketing
information
system or customer information system. There are then two types of data:
Contact
Data and Account Data. The Contact Data is shown below in Table 1. This
information is associated with an individual (or company) and the associated
address. Account Data is information that comes from the accounting system and
26
CA 02374120 2001-11-09
WO 00/68860 PCT/C1S00/13132
contains information such as the number of units, the dollar value, interest
charged,
date of sale, etc. If the information comes from prospects or outside lists,
it may
contain demographics, marketing campaign, acquisition date, etc.
Table 1
Contact Type Data
Data Element
Name Organization or individual
Supplementary Mailing Information
Address Line 1 through 3
City
State
Postal Code
Country
County
Account Number
Email Address Work
National Identification Number
Passport Number
Home Telephone Number
Work Telephone Number
Tax Identification or National Insurance Number
Email Address Home
27
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
Alternative Email Address
Link to Accounts (1 through N)
Dun & Bradstreet Numbers
Establishment
Headquarters
Parent
Ultimate
Census Code of Address
Geo Demographic Code
Account Representative
Organization Code
Holding Company
Organization Number
Branch Number
Step 1 B Split (into Contact and Account Data)
The account data is split away from the above information and contains the
account
number and information associated with the purchase of goods and services such
as
the date of purchase, item purchased, quantity, cost per item, and value of
this
purchase.
28
CA 02374120 2001-11-09
WO 00/68860 PCT/C1S00/13132
Step 1 C Scrub
Using conventional techniques such as Innovative DictionaryT"~ from Innovative
Systems contact data is parsed into words, patterns of words are identified,
individual data elements are identified, and lines are typed by (where
possible):
N Name Lines
O Organization Name
S Address Supplement Line
A Care Of Line/Attention Line
S Street Address
R Rural Address/ Box Number/
Mail Stop
C City/State/Postal Code
P Postal Area
L Country
K County or Geographic Area
In addition to the above-standardized information, the input lines are also
carried for
each customer. The system is multi-lingual with definitions and patterns and
line
orders for each unique country.
Step 1 D Load Account (Data)*
Account data extracted from the legacy systems is staged into a relational
database
for further processing. This data is organized by account number.
29
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
Step 1 E. Load Client (Data)*
Contact data extracted from the legacy systems is staged into a relational
database
for further processing. See the following sample tables.
*NOTE: The database was designed by developing business process models,
logical models, and physical models that organize and locate the data in the
desired
manner. The content, procedures and attributes used in these models are
proprietary to Innovative Systems, Inc. but were developed by conventional
modeling
and programming techniques utilizing languages such as Cobol and C.
Sample Input & Output Records
The below table shows the input and output record for the Contact portion of
the
Account data. Customer records are developed that link to the Account record.
Two customer records are created: one for George Jones and one for Sally
Smith. A
customer record is created for MegaCorp also since this company's name is
found
on the account record involved. Existing technology assumes that the customer
on
line 1 makes the purchase of goods and services and subsequent customers are
ignored. While this allows for all the customer data to foot to accounting
data, it
does not give any credit to other customers on the account or their having any
influence over the account. In the below example, three customer records would
be
created. Telephone numbers, tax ID numbers and like data that is not
associated
with a customer will be associated with only line one.
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
z
0
N
N
a
a
Y
U
a
w Q
O o~C
U
o
J O a
O ~ Q,
~ n
O J
a O ~
H
ul J O
N ~ N
tn ao m a ~1
4! a~D liJ ~ c0 (h C9 C'~ D a
"~ O ~ ''~ 'o ~ ~ D
c0 N d_~ ~ N
D O C3 ~ O v <t r C7 I- UU
Z Z O tn U
a~
a
E
o z
U
C
a
a
J O M
y Z E ~
= w Z ~ ~ .N N
J a a o O ~ = Z d
Z H L~tJY .iC N Z o .
O ~ U E ~ ~ E ~ o o
w c4 a ~ ~ N ~ s L m m
2 ~ z ~ v Z d y w =
o - o Q o a ~ f."~ 'm
w ,' ~
C n ~ ,a-JJ v E m c'~oo cxew
cn a m z a z ~ ~ ~
m
Q
m
c
'E
w
a
31
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
M
N
L
CMOc0 ~ N
Q n N N N O
INN O ~ N
D O O O n ~ ~ -~ O O N
d v
E r O C o
L (p m f0 N L m
f/1 7 a E o (p N
E .' E ~ E
I _ z o~
~ ~ U O z m z
Z
L
o E ~ o >
v
-o U m
C N
C N U O d U
U ~ O p N _O
m U
Q E ~ ~ v
0
o ~ ~ p ~ c U
~n o ~c U
U
D U C'3 Q 0 cn
32
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
n
L
Z
J
O
N
~_
Q
a
Y
U
Q
w O
U
Q d
d
C'3~ ~c~n
O
U O
a U
O J
U ~ Q
H
Z ~ ~ M M ~ D a
W
Q ,NZ N c~~Dm v Q ~ n
CDO _C'~ D lLQ
O O U' O d' ~ .- U' I-O
N
U
C
ft1
N
C
cC
C
Z O t0 U o
Y ,gyp ~ Z
o ~ Z d t
H O ~ '~
N c Z ~ ~ j Z
1 ~ Q p ~
Z o ~~ Z Z C ~ m
O ~ Y Z ~ U E C N O E
w d ~ ~ t o v n E 7
O ~ Z J Z ~ ~ Z ~ _a~C - ~
w ~ U
O U O ~ c Q m o H ' a - Q
C'3 Z Z . a~ ~ Q c
o~w ~' w v E iou o c"cZ .E ' c
n n d Q llJZ d S ~ H 111Q J
33
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
N
M
CO N
O
N
O ~ O c0
c0
N N O ~
(n
d O O N
N
N
_o E
E r ~ O c
L Q7 f~N Z
d v cS1N L
m E a ~ ~ c
._ Z E o,E
J I ~ U O Z
E ~ o m
~ U
~ a~o
m o m
m o
v ~ ' ~
m U o Q _ v
E .m
o 0
~
D d v U
m ' V
U C'3Q O ~n
34
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
z
N ~'~ - N
r r
a a
U : Y
_~.._,.=; U
Q
I Y
r = a ~~-a z a
Q
O o a ~ac9 o a
Q pOp J W pOp
W J
h (n
Iw
Z Z
W ~ W
d
W a~ i lL a
o
Ct O . D O
Z t4 U O tn U
O
Y Y
U U
O ~' O
~ ~
H ~ ~~ H ~
_
H w Q ~" ~ Q
a~ ~ ~ d
o ~ a
e e ~ e~ a
n n
a
z ~ Z
C OC r H~ C OC ~
'3 '3
Q _ a ~ ~ ~ _ a
= :~:= =
a y,.
tn a
o'
0 0 ~ z o .r o a z o
C~
U
w '
n cn a Q .._ ~ ~ ~ a Q
~
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
m m o
~
O ~C tN~ ~NCJI~ N N _
N N
J O O O O O O O
O O O O O
U N
L
O O > >
O N O O U U U
U V V V V
J a a a a a a a
O _X
(tf 7
U
-
Q d
'C3 Z I ~ (n
N F- Z Z
J fn O O n
X
W n ~ O ~ O ~ u.
O
_
(~ E ui m
a a a a J
- ~ ~
Q J C7 ~ O C7 OCO
. c~~ QUU ~ ~ C n
U G1
O a
H _ O _ O _
U ac
d
..- c
J N M r M ~ N
d
Q N M p M p N
O p p p Z O
O O O p O O O
O O O O O
L O O
O O O O ~J
N
N
O
_
U
N N d d
H O O O O O O Y
w
~ ~ U U U U U U
X
N
_
Q '
c0 z m ~ Z Z
N Z uj n
N ~ Z O
07 J O O n O
~-.
V
N
L
L X
~ ~ ~ ~ ~ O O ~
47 N
- a a
0 E ium O O w
>- ~ a a ~
O w w a a w w w
~. i c4c9 cn cn ~ ~ cg
a
H _ _ _ _ O O _
d
U J r ~ N N M M r
U
' d O O p ~ M M O
U O O O O O O O
: ' o o o o o
~ o o o o o o o
O~ ~ d
i~- ~ ~ 0 0 0 0 0 0 0
J J U U > > > > >
U U U U U
36
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
°o °o °o o °o o °o
C c c c c
> > > > > >
U U U U U U U
Q Q Q Q Q Q Q
U
M
c~ co a> v d 0
O ~ ~ d' N ~ C~ J H O O D
C C O C U ~ C
L L .iC O
j d d U .I4 'gyp ~ V z
N N
U O ~C z U a _>
N
N d ~ Y ~ ~ N ~ - N C O ~ O
E E o E x N ~ ~ E u'~ -o ~c r
a z z ~ z ~ o S z w i a m
w w w w w w w
z z z z z z z
O O O O O O O
w w w w w w w
O c~ c~ c~ O c~ O
Q
O O O O O O O
m m m m m
c~ c~ c~ c~ O c~
c~
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
a~ m d a~ a~
a~
E E E E E E E
0 0 0 0 0 0 0
M (A (n (A N N (d
7 ~ 7 7 ~ 7
U U U U U U U
37
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
Step 2A: Matching Individual Names
This process matches First Name, Last Name, Name Suffix, Gender Code, Street
Number, Street Name, Postal Code, and Individual identifiers such as Social
Security Number if available. There is a variety of software available in the
marketplace to perform this task such as Innovative MatchT~~ from Innovative
Systems.
Step 2B: Matching Organization Names
This is the same process as 2A, except for organizational names. There are
products available that specialize in corporate name duplicate identification
such as
Innovative CorpMatchT"~ from Innovative Systems Inc. Records to be collapsed
(combined) to unique organization names at a unique address will be
identified. "No
Link" cross-references are posted for records that should not be linked. In
addition,
if two records link, but the names do not match exactly, an alias search key
is
generated for the search algorithm that is connected to the database.
Step 2C: Combine Duplicates
This process combines customer records that are duplicates. This entails
having
one customer survive with a combined account and customer cross-reference
list.
On the database, records will be combined that are considered to be the same
customer. Combine is defined as moving the cross-reference and associated data
from one customer to the surviving customer. This process puts the non-
surviving
38
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
record in a delete status and creates alias records for searching where there
is a
name variation between the two customers being combined. In addition, records,
which come together in the matching process which, are not duplicates are
stored on
the customer to application list for identification by the "No-Match" module
of the
software that executes this process of the invention. For example if customer
A
does not link to customer F, but the matching software brings them together,
then
under customer A's list, there would be a "No-Match" pointing to customer F.
Under
customer F, there would be a "No-Match" pointing to customer A. Thus in
subsequent runs, the matching software available from Innovative Systems would
cause the software not to link these two customers. This is a laborsaving
device that
eliminates future review of link results. In addition, if two records link,
but the names
are at variance, an alias search key is generated for the search algorithm
that is
connected to the database. In all cross references, creation date and delete
date
are kept in this and future steps. Cross references are never created. This
allows
buying behavior to be analyzed with consistent customer and household
definitions
by using the composition in one period, but the purchasing behavior in
multiple time
periods.
Step 2D: Extract Customers
Separates records into individual and corporate name and address records.
Step 2E: Household
This process is similar to the process in 2A, except that the first name is
not required
as matching criteria.
39
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
For use of the system in Spanish and Portuguese speaking countries, a process
known as "Spanish Match and Household" is used to determine duplicates when
there are double surnames. Assume there are two last names in each customer
record and they are in last name field one and last name field two.
If there is customer record A and customer record B, then:
- field 1 record A can link to field 2 record B;
- field 2 record A can link to field 1 record B;
- field 1 record A can link to field 1 record B; or
- field 2 record A can link to field 2 record B.
Any combinations of these linkages are an indication that they are from the
same
family if they are at the same address. If the first names are equal, they are
more
than likely duplicates.
Again, if records are placed in a household that do not belong there, a "No-
Match"
record is generated that prevents the records from being assigned to the same
household in future runs. The date of household formation is generated for new
households along with the first customer date. Where households split or
combine,
household numbefs that survive are the ones 'for most remaining members. New
household numbers are given to the new formation. The old household numbers
are
saved a delete code and date of the delete. This allows tracking household
composition over time periods and connects families together even after one of
the
children has moved (or the subsidiary is sold in the corporate household
case).
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
Step 2F: Corporate Household
The Contact record is linked to a third-party file containing corporate
hierarchy data,
such as Dun & Bradstreet. A list like the one shown below is constructed.
Sample Table Showing Customers Linked with Company Data
List
with
Dun
&
Bradstreet
Link
Customer0000000033 MEGA CORP Establishment 15643 Duns
Customer0000000033 MEGA CORP Headquarters 65423 Duns
Customer0000000033 MEGA CORP Parent 79243 Duns
Customer0000000033 MEGA CORP Ultimate 88982 Duns
Step 2G: Post Households
In this process, the household number is posted to the customer list. After
the initial
run, if there is an existing household, the old household number is kept for
referential
history. All referential history is time dated with the creation date and
reassignment
or deletion date (if they occur). This permits a marriage union (or corporate
merger)
to point back to the families or entities that were the source of the union.
This is very
important in tracking wealth formation and corporate hierarchy evolution or
merger
history.
Step 2H: Extract
Extracts names and addresses and the social security number for records that
are in
the same Household using the database of Marketing Customers.
41
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
Step 21: Real CustomerT"' Linking (Female Matching)
Records that have the same female first name at the same address and have the
same social security number are considered to be the same customer. The
customer record that matches other records in a household last name for a
multi-
member household is considered to be the true name. If this is a single member
household, the name with the most applications is chosen. A record is added to
the
list, which shows the account type as being the True CustomerT"' for the
record that
is considered to be a member of the Real Customer grouping, but both are left
on
the database. The different first and last name is then posted to the "alias
locate"
file. For example, a person may use Sue Jones at work (maiden name) and Susan
Dewey at home. Both names might be on two accounts at the same address.
Step 2J: Post Real Customer Links
Where two Marketing Customers are considered to be the same, one customer has
a relationship code in its cross-reference list of Real Customer with the
other
Marketing Customer's identifier. An alias locate record is generated in this
case.
Step 2K: Extract
For each Marketing Customer this step extracts its Marketing Household and
Real
Customer cross-reference (if one exists). Accounts with more than one customer
also have their customer-to-account cross-reference extracted.
Step 2L: Real HouseholdT"' Linking
42
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
Individual records that are sourced from the same account are assumed to be
members of the same households. For example, if there is a mother-in-law
living at
the same address on the same account, this person is assumed to be a member of
the household. This will handle children of the first marriage where the
mother
remarries also. The first household is called a Marketing Household. Under the
Marketing Household cross-reference, a Real Household cross-reference is
created
and maintained in the database. This allows the market researcher to get a
much
better idea of the actual number of households. Whereas not employing this
technique will lead to over counting households and undercounting cross-sell
ratios,
the opposite occurs in this case. There will be mothers-in-law who share an
account,
but do not live at the same address and are really in separate households.
There
may be some undercounting of households, but it will be less than the over
counting
by not using this technique.
Step 2M: Post Real Household
This program ports the Real Household to the Marketing Customer cross-
reference.
Step 2N: Extract Social Security and Customer
Extracts the name, address, and social security number of each Marketing
Customer.
Step 20: True Customer Linking
43
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
The social security numbers (or email or phone number or passport number, if
the
user so desires), first, last name, and suffix are compared from the list file
generated
from Step 2N. If they are equal (or close enough under the user's definition),
the
customers are assumed to be the same person.
Step 2P: Post True Customer
This program posts the True Customer cross-reference to the Marketing Customer
cross-reference and date of the update (or delete). This program is used
repeatedly
in the process to post all cross references generated (outputs from 6D and 6F
for
example).
Step 2Q: Extract Existing Household and Customer Cross-References
Extracts the Real Household, Real Customer, True Customer and Marketing
Customer cross-references generated from Steps 2A through 2P and Procedure 1
above.
Step 2R: True HouseholdT"" Linking
If the True Customer falls into two Real Households or two Marketing
Households,
then the households are combined to form a True Household.
Step 2S: Post True Households
True Households are posted to the cross-reference list for associated
Marketing
Customers and date of the update(or delete).
44
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
Step 2T: Extract All Cross-References Associated With Marketing Customer
Extracts all cross-references described above.
Step 2U: SuperHouseholdT"' or Social NetworkT"' Linking
SuperHouseholdsT"' are considered to be any organization or individual that
shares a
common link. Constructing a list of the SuperHousehold with all the customer-
to-
customer linkages allows the presentation of diagrams of the type shown below.
An
Account Type is maintained for each type of shared link .
Sample Connection Types:
~ Another Customer ~ Joint Account Number
~ Account Numbers and Account Types ~ Household Number
~ Telephone Number ~ SuperHouseholdT"~ Number
~ Alias (another name that is used) ~ Link to Account Number
~ Social Security Number ~ "No Link" Customer
~ Email Address ~ "No Link" Household
~ Dun & Bradstreet Number ~ Hard-Link Customer
~ Real CustomerT"" ~ Hard-Link Household
~ Real HouseholdT"" ~ True CustomerT""
~ True HouseholdT""
Step 2V: Post SuperHousehold or Social Network links
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
The date of the update or delete for the cross reference is kept when the
super-
households record is posted to the marketing cross reference list.
Procedure 3 - Incremental Adds, Chancres and Deletes
This procedure allows the Account contacts to be compared to the record
generated
in the last time period (see Figure 9). Alternatively, this maintenance may
trapped in
real time and sent to the update process as they occur. Three possible record
types
are identified: Adds (account records that are new), Deletes (account records
that
are no longer there or are closed), and Changes. Alternatively, if maintenance
from
disparate systems can be trapped real time, the maintenance data may be sent
to
the incremental update process real time. The records are scrubbed as
discussed in
the section on scrubbing. The Deletes are categorized as moved, moved with no
forwarding address, or account deleted or closed. Once scrubbed, all records
are
sent to the database through an incremental update process.
Procedure 4 - Post and Distribute
New records are incrementally matched to the database by comparing against key
fields and evaluating the new Adds against the existing database.
Methodologies
such as Innovative-FindT"' and Innovative-UpdateT"" incrementally update
client
databases of this type. The content, procedures and attributes used in these
systems are proprietary to Innovative Systems, Inc. but have been developed
using
conventional programming techniques and languages such as Cobol and C. If
there
46
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
is a match, then a cross-reference is posted to the customer record and an Add
is
made to the account file.
If there is a change of address, the address change is posted to the customer
record. The DistributorT"' function determines if the individual has other
accounts.
Depending on the rule set, update records are sent back to the legacy systems,
including other CIF's and data warehouses. Otherwise, the new address creates
a
new customer where the old customer record and the new customer share a "True
Customer" key. A search is also made of the database for any records that
match
the customer at the new address. Incremental Householding is also done. If the
customer joins an existing household, a reference link is created to the
household (to
maintain referential history) that the customer came from with a like-link
created for
all the members of his former household. If the record causes a new household,
the
referential history is kept.showing the old household the person belonged to.
Thus, if
a child marries into another family on the database, when the two children
leave their
respective households, time-dated referential history is available so that
both are
shown in the networks that they came from. This is very useful in tracking
wealth
formation in the financial services industry. In addition, this permits
comparisons of
buying patterns between two time periods with the same household composition.
Procedure 5 -- Attribute Account Information to Marketing Customer
(Existing State of the Art)
This procedure sets up all of the links to account information and other
customers for
each Marketing Customer record generated in Procedures 1 and 2? (see Figure
10).
This procedure can be accomplished using conventional techniques as contained
in
software such as Innovative - MkIST"" Marketing System
47
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
Step 5A: Extract Client Account Cross-Reference
As discussed previously, a list is developed for every type of relationship on
the
database that involves a customer (Marketing Customer). The list contains the
customer identifier and a link field. The link fields are:
~ Another Customer ~ Joint Account Number
~ Account Numbers and Account Types ~ Household Number
~ Telephone Number ~ SuperHouseholdT"' Number
~ Alias (another name that is used) ~ Link to Account Number
~ Social Security Number ~ "No Link" Customer
~ Email Address ~ "No Link" Household
~ Dun & Bradstreet Number ~ Hard-Link Customer
~ Real CustomerT"' ~ Hard-Link Household
~ Real HouseholdT"~ ~ True CustomerT"'
~ True HouseholdT""
For the customer, the line number and the customer order number within the
line is
kept for each customer. A cross-reference file is created of all customer
account
number records found of the first customer found on line one.
For example if line one had Bob & Mary Jones with Account # 524321, a record
would be created for Bob Jones showing a link to the account. The record would
show that Bob Jones is linked to Account # 524321. In addition, the Marketing
Household number is placed on this cross-reference record which provides for
Bob
48
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
and Mary to be in the same household. By definition, a Marketing Customer may
belong to only one Marketing Household.
Hard-Links are defined as customer records that should be combined but are not
recognized as the same customer with matching software. These are placed in
the
file by manually reviewing match and household reports, and alias records are
automatically generated for the search function. Similarly, records that
belong in the
same household may be linked together with a Hard-Link function. This forces
them
to stay in the same household, regardless of what happens in the Householding
process. "No Link" functions are also posted. If in the customer and household
matching function clients are found to potentially link, but in fact do not
link, a "No
Link" customer or household cross-reference is generated. This ensures that
records previously determined not to link remain unlinked.
Step 5B: Marry Account to Customer File with Account Data
Account systems typically house data on sales dollars, units sold, current
balance,
etc. (buying behavior or customer or account attribute). The system may use
any
one of these values to develop the measure of Influence Value the user
desires.
The Customer Key to Account Key cross-reference file is sorted by Account Key.
There may be multiple Account types, and this process is done for each Account
type. Account attributes (account number, dollar sales, units, balance, etc.)
are
attached to a customer key using the cross-reference table. Only one customer
key
is attached to each account type/account number record with the user selected
buying behavior or customer or account attribute.
49
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
Step 5C: Aggregation Development of Balance Records for Marketing
Customers and Marketing Households
The resultant file (from 5B) is then sorted by household key and then by
customer
key, account type, and account number. A summation is then done for each
customer and household by account attribute (buying behavior) that is to be
analyzed. Once customer account records are aggregated by marketing and
customer attribute, they are then sorted on the summed attribute from the
highest to
the lowest and ranked from 1 to n, where n are the number of customers (or
households if the household balance record is being developed). N is then
divided
into the rank to compute customer percentile, based upon the relationship
value or
buying behavior statistic chosen. The process is repeated to determine
household
ranks and percentile within the Household marketing attribute database
Procedures 6 through 8 -- Familial Collapse
Since only one account attribute goes to one customer record, there is audit
control
that allows the sum of customer attributes to equal the enterprise attribute
total (units
sold, sales footing, deposit balance, etc.). In addition, there are Marketing
Customer
and Marketing Household counts. Procedures 6 through 8 are designed to arrive
at
this aggregation of customer attributes for each enterprise attribute to be
analyzed.
Procedure 6 accomplishes this for the True Customer and Real Customer (see
Figure 10) while Procedures 7 and 8 (see Figures 11 and 12, respectively)
repeat the
steps used in Procedure 6 for the Real Household and True Household,
respectively.
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
Procedure 6: Familial Collapse Part I
(for True Customers)
Step 6A: Develop True Customers
The file developed in Procedure 5 of the cross-references of Marketing
Customer to
True Customer, Real Customer, True Household, and Real Household is sorted by
Marketing Customer and True Customer. The True Customer ID replaces the
Marketing Customer ID.
Step 6B: Sort by True Customer
The file is then sorted by True Customer ID.
Step 6C: Aggregation
The aggregation process is repeated from Procedure 5 resulting in the count of
True
Customers and True Customer balance records.
Step 6D: Reset Edges
The customer-to-customer file is sorted by marketing and True Customer. The
True
Customer identifier then replaces the Marketing Customer in the "from" column
and
resorted on the "from" customer identifier. The process must be repeated for
the "to"
cross-reference. The identifiers are then reset to edges for pictorial
representation
of True Customers network.
51
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
Steps 6E through 6H: Repeat for Real Customer:
Steps 6A through 6D are repeated for each Real Customer.
Conclusion:
For targeting and analysis, databases now exist for both True and Real
Customers.
Procedure 7 - Familial Collapse - Part II
(Create Real Household Balance Record and Edges)
An extract of Marketing Customer to Real Household is extracted. If there is
no Real
Household belonging to a customer, the Marketing Household is extracted. For
customers with Real Households, the Marketing Household key is replaced and
the
data is aggregated again.
Procedure 8 -- Familial Collapse - Part III
(Repeat Procedure 7 for True Household)
Procedure 7 is repeated to develop the edges for the True Household.
Procedure 9 -- Influence Value
This step describes the process of valuing a customer for his, her or its own
worth as
well as determining the value of the influence that customer has over other
customers, which provides the user with the ability to alter the way sales and
marketing channels are configured and balanced. This process is accomplished
by
52
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
examining customer relationships through "Social Networking", a technique
developed by sociologists and communications specialists to illustrate the
communications between individuals. Figures 15 and 16 show a diagram of the
communication patterns in a company. This relates to customer management
through the diagram shown in Figure 3, which shows that the CEO of this
company
has a family and a variety of business interests. This has ramifications on
the way a
company doing business with Mr. Smith will handle his account, and the way
that
company measures the value of all the additional business that Mr. Smith has
influence over bringing to the company.
In the example of a financial services company such as a bank, this has a
great
implication on how sales channels are configured and how pricing is done.
Banks
have long recognized this phenomenon and put into place departments that are
meant to cater to individuals of this nature (private banking). The difficulty
is that
these departments have had no way to determine which customers belong in this
department except on the value of their individual relationship. The table
below
shows how much business each customer does with the bank.
Customer Household Relationship if
Balances Total Total John Smith
John Smith CEO
Checking $1,351 $1,351 Owner
Linda Smith CFO Wife
Statement Savings $14,652 Owner
IRA $24,758 Owner
Pension Account $120,600 Owner
53
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
Trust Account $842,000 Owner
Brokerage $586,000 Joint with Sally
$1,589,361
George Smith Child
Trust $35,000 Trustee
Savings $2,480 Guardian
$37,480
Annie Smith Child
Trust $15,000 Trustee
Savings $654 Guardian
$15,654
Ben Smith Child
Trust $8,285 Trustee
Savings $120 Guardian
$8,405
Mariann Smith Child
Trust $6,984 Trustee
Savings $475 Guardian
$7,459 $1,659,710
Sally Jones Mother in Law
Checking $8,295 Joint
Checking $758 None
Brokerage $74,325 None
$83,378 $83,378
Mega Manufacturing Owner with Wife
Checking $59,640 Signature
Pension Plan $1,752,986 Trustee
Foreign Deposit Pounds$55,982 Signature
$1,868,608$1,868,608
Total Network Value $3,611,696
On the basis of the from information available through
above the traditional
marketing systems,
as a customer John
would not be eligible
for private banking
as
54
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
he has only $1,351 in deposits. Under normal circumstances, John could be
charged for check processing instead of having his fees waived, an action that
could
potentially damage his relationship with the bank, and thus the ability of the
bank to
reap the benefits of additional business or prospective business John has
influence
over. Additionally, numerous account managers would be assigned to John; for
example, individuals from the following departments: trust (one for
investments and
one for administration), middle market corporate lending, retail banking, an
officer in
the London office, and real estate lending. His wife could also have the same
number of lenders and also have a manager from the trust department on the
pension plan and 401 (k) plan. Potentially, the number of bank personnel that
handle
this relationship could be reduced. In addition, one department could
understand
how their actions affect those of another department.
Examining the table shows that as a customer John is worth $1,351 in deposits.
He
is a member of a household that has deposits and investments of $1,659,710. He
is
on a joint account with his mother-in-law, and his mother-in-law (even though
she
lives with him) has investments and deposits of $83,378. The family's
privately held
company has deposits and investments of $1,868,608. The sum total of all the
relationships that John can influence is $3,611,696, which would not otherwise
be
identified absent the calculation of an "influence value" as in the present
invention. A
measure of how organizations fail to identify this customer attribute may be
derived
by ranking customers on their sole value and then on their influence value.
For
example, John may be the 1000t" customer in ranking based on his deposits.
Based
on his influence, he is 10t" . Subtracting his customer ranking from his
influence
ranking can highlight the degree of under measurement of a customer's value.
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
"Influence value" is defined as all the business (sales, deposits, profits,
loans, etc.)
that is derived by summing the metrics of the customers one or more nodes away
from an entity such as an individual or company, and can be measured using the
following procedure (see Figure 13):
Step: 9A: Extract (Customer Cross-References)
As discussed previously, a list is developed for every type of relationship on
the
database that involves a customer (Marketing Customer). The list contains the
customer identifier and a link field. The link fields are:
~ Another Customer ~ Joint Account Number
~ Account Numbers and Account ~ Household Number
Types ~ SuperHouseholdT"' Number
~ Telephone Number ~ Link to Account Number
~ Alias (another name that is used) ~ "No Link" Customer
~ Social Security Number ~ "No Link" Household
~ Email Address ~ Hard-Link Customer
~ Dun & Bradstreet Number ~ Hard-Link Household
~ Real CustomerT"' ~ True CustomerT"'
~ Real HouseholdT"' ~ True HouseholdT"~
For each customer, one record is created for each customer on the database and
the customer that they link to.
56
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
Step 9B: Marry With the Balance Records
The customer-to-customer cross-reference file is sorted by customer
identification
number and account type. In Process 5, a balance record for each Marketing
Customer was created. The balance record is attached to all linked customers.
Example:
Customer Type Link to CustomerType Customer A Balance
A I A ( $1,000
A I B I $1,000
A I C O $1,000
A I D A $1,000
A I E I $1,000
A I F A $1,000
A I G I $1,000
A I G I $1,000
[Note: In the "Type" column, "I" represents an individual; "O" represents an
organization; and "A" represents an Affinity relationship to an organization
as defined
previously]
In the above example, customer A has a balance of $1,000 in all of his
accounts
combined. The first record that is created shows customer A linking to itself.
The
type shown is derived from the definition in Process 5A.
57
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
Step 9C: Develop Influence ValueT""
The file created in 6B is then sorted by the "Link to Customer" field, and
then by the
"Customer" field. The occurrence of a customer number within similar "Link to
Customer" groups is unique.
Three types of records are developed, one for each record type. Thus, there
are
summary records for like record types and a maximum of three cross types
(Examples: individual to organization, individual to affinity, organization to
individual,
organization to affinity, organization to organization, individual to
individual). Two
summary records are created: one for direct influence (not Affinity
Relationships)
and one for Indirect RelationshipsT"' (Affinity Relationships).
Thus, through the development of these records, a measure of influence a
customer
has directly and indirectly on customers to whom there is a relationship
within the
database is created.
Procedure 10 -- Customer Reportinct & Analysis
Using the balance files for influence and customer balance, the links may be
shown
to other customers for a selected customer. In addition, the influence the
customer
has on other customers may be shown (see Figures 5 and 6).
Example:
John Jones
Customer Balance $10,500 percentile 85.64%
58
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
Household Balance $32,000 percentile 88.43%
Influence
Individual to Organization $2,562,000 percentile 15.4%
Individual to Individual $120,000 percentile 8.42%
Total Influence Direct $3,682,000 percentile 9.42%
Influence Affinity
Individual to Affinity $1,523,422 percentile 10.52%
Procedure 11 -- Analyze Data
The resultant information is then used to create sales, credit, and marketing
intelligence using commercially available software.
End of Procedure description.
This invention may be extended from a business to consumer or business to
business sales, marketing and customer management to any problem where
networks of relationships might exist. One example of such an extension is in
the
area of pharmacology.
59
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
For example, diseases may be categorized by:
~ Family
~ Morphology (the characteristics under a microscope)
~ What is affected:
- Organ
- Tissue
- Joint
- Bone
~ Symptoms
Drugs may be categorized by:
~ Compound
~ Disease treated
Using the system of the invention it is possible to build lists of all drugs
that have
common compounds. Compounds can be isolated, then, that are common to the
treatment of a symptom or symptoms, family of disease, or a specific disease.
Further, the analogue of influence may be extended to this application by
giving the
success rate of a drug, imputing the success rate to the compound, and then
adding
up the joint success rates of compounds. Potentially, then, this could lead to
drugs
developed from combinations of compounds found in existing drugs. At a
minimum,
cross-references are built of compounds, which are cross-referenced to drugs,
disease family, and microscopic characteristic.
s0
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
Figure 1 shows a compound that occurs in two drugs. Figure 2 shows the family
relationship being collapsed to diseases. This shows that in Disease 2, both
Drugs B
and C are effective and they both share a common compound. If Drug B and E
shared a compound and were effective on Disease 2, then possibly a new
protocol
could be developed using the compounds in common in Drugs B and C with the
compound in common with B and E. Further, the compound that is common to the
most number of drugs in a disease family might be considered to have the most
influence. Variants on quantity of compound and disease category may also be
examined.
While presently preferred embodiments of the invention have been shown and
described in particularity, the invention may be otherwise embodied within the
scope
of the appended claims.
61
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
APPENDIX A: DEFINITIONS
The words "process" and "system" are used interchangeably.
Marketing Client (Customer)T"": All accounts that share a common last name,
first name, address, and tax identification number (if present).
Automatic alias generation: When two customer records are combined and
first and last names have differences, locate record using the first and last
name
that does not survive on the database.
Direct RelationshipT"': Two direct customers sharing an account, phone
number, tax number, D&B number, or email address where the organization
comes from line one only.
Edge: A relationship between two customers (or households) which shows both
identifiers in a list.
Equal: May be defined by the user as a misspelling or mismatch that is
acceptable.
62
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
Indirect (Affinity) RelationshipsT"": Relationships as defined in Direct
Relationships where an organization occurs on line 2 or greater of an
accounting
application.
Marketing HouseholdT"": All customers with the same address and last name.
No LinkT"": Pairs of customer or household keys that indicate two customers
should never be linked or a customer should never be put in a household.
No-Match LinksT"": In the application cross-reference list, customers and
household cross-references that are not permitted are kept.
Real CustomerT"": a) A female first name equal, address equal, and social
security equal; b) Male with a double surname in one record that has an equal
first name; c) Companies at the same address with the same tax number and/or
the same D&B establishment number and/or headquarters number.
Real HouseholdT"": Customers with joint accounts and different last names, and
also Real Customers with one or more Marketing Customer records.
True CustomerT"': a) Customers with equal first and last names and equal tax
numbers and at least one last name at the second address are equal to one last
name at the first address for multi-member households; b) Organizations with
same tax number and/or headquarters or establishment D&B number.
63
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
True HouseholdT"": Bringing the True Customer relationships into the Real
Household.
Familial CollapseT"": A system to re-cluster data by household or by combining
links that are one or more links away (each link away is called a Degree of
Influence).
Time-Dated Referential History: All previous links of Marketing Customers are
kept. This allows the household composition for the measurement of "buying
behavior" to be the same over two time periods and eliminates behavior that is
the
result of household shifts and not buying behavior.
Influence ValueT"": Categorized as direct and indirect. Direct influence is
categorized as individual-to-organization, individual-to-individual,
organization-to-
individual, and organization-to-organization. All customers are categorized as
individual or organization. Individual-to-individual adds the measures of all
related
individual customers together that are related to one customer. If the
individual is
related to an organization customer from line one of an account, then the
individual
is said to have a Direct Relationship. If the individual is related to an
organization
from line 2 or greater, the influence is considered to be indirect or
affinity.
64
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
Incremental MatchT"": The process of matching incremental updates to names and
addresses by joining cross-reference lists of fragments of data that do not
require
the off-loading of all names and addresses.
DistributorT"": The process of determining which applications receive name and
address updates if one application name and address is changed for a customer
that
has many applications.
Spanish Match and Household: The process of determining duplicates when
there are double surnames. Assume there are two last names in each customer
record and they are in last name field one and last name field two. If there
is
customer record A and customer record B, then:
- field 1 record A can link to field 2 record B;
- field 2 record A can link to field 1 record B;
- field 1 record A can link to field 1 record B; or
- field 2 record A can link to field 2 record B.
Any combinations of these linkages are an indication that they are from the
same
family if they are at the same address. If the first names are equal, they are
more
than likely duplicates.
Affinity OrganizationT"": Organization line on line 2 or greater in the
original
application accounting record.
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
Customers and Households Real and True: (see definitions).
Appendix B: Example Proctrammina Architecture
The system of the present invention can be implemented in both a batch and API
format. The batch version allows for the creation of social network linkages
across
the whole of a dataset to whatever depth (or distance) as the links will
sustain. In
other words, there is no means for limiting the distance of relationships it
creates.
The API version provides the capability to tune the creation of links far more
extensively than the batch version and presumes that a relational database
management system (RDBMS) is in place.
Batch
The batch version of the system consists of two parts, a Windows GUI program
and
a series of eight C programs. Some of the C programs are run multiple times
creating 15 steps in the process. All of these steps are executed through a PC
DOS
BATCH file, script file or procedure depending on the operating system. The C
programs can run on any platform supporting the C language. The GUI is
currently
limited to the Microsoft Windows environment. The C programs receive their
operational information from a common INI file. They all read in the INI file
as a
command line argument. The INI includes things such as where the input file is
located, where the output file can be written to, and the lengths of the
records and
fields within the records. This INI file is a text file that can be made
available on any
66
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
platform where the C programs are to be executed. The purpose of the GUI
interface is to build the INI file in a user-friendly fashion. All of the
details needed by
the C programs are presented by the GUI for the user to enter. The information
needed is as follows:
Name and location of the input file
Name and location of the log file to be produced
Name and location of the output file
Location of the C executable programs
Location of a file (referred to as a VFD file) describing the input file's
characteristics
The VFD file, which may be built with the GUI, aids in defining the data
needed from
the input file. The length of a record in the input file must be given. In
addition, the
length and displacement in the record of two fields must be given. One field
is the
KEY field, which is typically a customer identifier. This will become a node.
The
other field is typically a relationship. This will become an edge.
These edges are relationships such that two nodes share: telephone number,
SS/TIN, D&B numbers (establishment, parent, headquarters, and ultimate D&B
numbers; joint account; household; and referential history). Sets of links are
used to
build lists of edges. The lists are then joined to form a network.
67
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
Once the INI file has been defined the running of the C programs can be done
through the GUI or the INI can be saved and used outside of the GUI. Although
not
available to be changed through the GUI the INI also contains all of the
labels,
messages, warnings used in the system. These are currently in English but
could be
changed to another language if needed.
After the system has been run the output file contains the KEY field, the KEY
field of
a 'connected to' record and a clustering number common to all records that
have
been married together. This file can then be loaded into a database or other
repository and viewed through a display mechanism.
API
Overview:
This product will discover direct and indirect relationships between a series
of clients
(individuals and/or organizations). These relationships are stored in a
relational
database management system. Optionally, it will compute the influence, as
defined
here, each of these clients exerts on one another by examining account balance
information. The criteria used to build these relationships are user defined;
and they
may be used individually or in combinations. The user also has a means to
control
whether or not the results of this product are stored in the RDBMS for future
examination and the ability to limit the depth of the discovery. This product
may be
integrated into a third party graphical visualization tool to better
understand the
nature of the data. Currently, the product exists as a Windows 95 compatible
68
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
dynamic link library (DLL) written in ANSI C and embedded ANSI SQL working
with
an Oracle database.
Definition of terminology:
Node Any unique individual or organization within a network.
Edge The means by which two nodes are related.
Network A collection of nodes and edges; all edges are of the same type.
Social Network All notes that are connected by edges that form a set.
Direct Relationship Object A is related to object B. Object B is related to
object C.
Indirect Relationship From direct relationship, we infer that object A is
related to
object C.
Public Interface:
There are several publicly accessible function calls used to interact with the
product.
A list and brief description of each follow:
int chmlnit (short postResults, short depth, short influence );
This function initializes several flags used during the run of the product.
The first
parameter is used to determine if the resulting network should be stored in
the
database. The second parameter is used to limit the depth of the network. The
last
parameter determines if the influence computations are to be done.
int chmDestroy ( void );
This function is reserved for future use.
69
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
int chmlnitNetwork ( void **networkPtr );
This function's purpose is to initialize a memory location such that it is
suitable to
store a network. It accepts one parameter and returns success/failure. The
sole
parameter is a user-specified address in the computer's memory where internal
data
structures are stored and accessed.
int chmBuiIdNetwork (void *networkPtr, unsigned long id, ISI SQL_ERROR
*se);
This function's purpose is to build a network. It accepts three parameters and
returns
success/failure. The first is the address of the network. The second is a
unique
numeric identifier that maps to a single individual or organization in the
RDBMS.
The last parameter is a user-specified address that is used as the
communications
area between this product and the end user's. If an error occurs this function
returns failure and includes detailed information in the third parameter.
int chmListNodes ( void *networkPtr, DB ROIN * rowArray);
This function's purpose is to list all the nodes in a given network. It
accepts two
parameters and returns success/failure. The first parameter is the address of
the
network. The second is an address (user specified once again) of a buffer
which
contains a list of all nodes in the network.
int chmListAdjNodes ( void *networkPtr, long id, DB ROW * rowArray);
This function's purpose is to list all the nodes that are adjacent to a given
node in the
network. It accepts three parameters and returns success/failure. The first
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
parameter is the address of the network. The second is the identifier of a
parent
node. The third is an address of a buffer, which contains a list of all the
nodes in the
network that have a direct relationship to the parent node.
int chmDestroyNetwork( void *networkPtr);
This function's purpose is to clean up the internally allocated data
structures. It
accepts the address of the network and returns success/failure.
int chmGetNode (void *networkPtr, long cid, DB ROW * row );
This function's purpose is to return relevant data about a single node in the
network.
It accepts three parameters and returns success/failure. The first parameter
is the
address of the network. The second is the identifier of a single node. The
third is
an address of a buffer that contains various information about the node:
int chmConnect (char *connectStr, ISI SQL_ERROR *se);
This function's purpose is to connect to the database. It accepts two
parameters
and returns success/failure. The first parameter contains user and database
specific information, including a user id, password, and the name of the
database to
be used. The second will contain error information, if an error occurs.
~1
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
int chmDisconnect (!S1 SQL_ERROR *se);
This function's purpose is to disconnect from the database, it returns
success/failure
If an error occurs the sole parameter will contain detail information.
Parameter File Description:
The parameter file resides in the same location as the product. It is divided
into
logical blocks denoted by square brackets. Each block contains six fields:
active,
IinkDataType, IinkType, queryl, query2, and label. The active field may
contain a 0
or 1, which indicates that the product is to skip or process the block. The
IinkDataType field may contain a 0 or 1. A 0 in this field indicates character
based
data, while a 1 indicates numeric. The IinkType field contains a sequential
number,
to differentiate types of relationships. The queryl field contains a valid SQL
statement that accesses the database and returns one or more rows given a
client
identification number. The query2 field contains a valid SQL statement that
accesses the database and returns one or more rows given the results of
queryl.
The label field is used to simply describe the type of relationship you are
examining.
An example of a single block:
[ test ]
active=1
IinkDataType=0
IinkType=1
queryl=select ssn tin num from clnt core cc where cc.clnt_id = :b0 ;
query2=select cc.clnt_id, cc.nm_lin from clnt core cc where cc.ssn tin_num =
b0;
72
CA 02374120 2001-11-09
WO 00/68860 PCT/US00/13132
label= Network on Social Security Number
Output Description:
As stated earlier, the product optionally stores the network in the database
after it
has been built. A result table should be created prior to running the product.
The
table structure is as follows:
PARENT NOT NULL NUMBER(10)
MEMBER NOT NULL NUMBER(10)
KEYDATE NOT NULL DATE
KEYTYPE NOT NULL NUMBER(10)
KEY NOT NULL NUMBER(10)
The parent and member fields contain the unique identification number of a
single
individual or organization in the database. The keydate field contains the
date that
the network was built. The keytype field corresponds to the IinkType field
from the
parameter file, and is used to differentiate types of networks. The key field
is used
to easily group common members of a network together.
73
