Note: Descriptions are shown in the official language in which they were submitted.
CA 02258383 1999-O1-08
-1-
Improved System and Method and Articles of Manufacture for Automated Advisory
Decision
and Control Services Using Improved Decision Support Systems with Model
License
Protection
FIELD OF THE INVENTION
The present invention relates to automated and assisted decision making.
Automated
assisted decision making is a relatively new field made possible by the advent
and large-scale use of
powerful general purpose and special purpose processors, and advances in the
fields of decision
science, game and utility theories, and graphical user interface technologies.
Automated and assisted decision making may require initial input information
from
human experts and automated data collection systems in order to assist less
knowledgeable or time
restricted human operators or systems to make decisions. Such methods require
detailed
information gathered by knowledgeable experts in a particular field and
formulated by said experts
into a detailed knowledge base, where said knowledge base may be represented
by a model
consisting of criteria, and alternative strategies or items for selection, and
criteria weights,
evaluation information in the form of numeric values and textual and
multimedia meta-data, and
other such attributes as related to the criteria, the relationships amongst
said attributes, and
alternatives and alternative specific attributes and relationship of the said
alternative attributes with
the criteria attributes. Summarized and specifically tailored information,
including attributes and
attribute relationship, may be required by those less knowledgeable in the
field but who may need
to make a complex decision relying on said information, where said information
can be very large.
Assistance by means of human intervention and/or automated machine and
automated processes can
be applied to best tailor the complex information required for a specific
decision and make the
decision manageable and well informed. The detailed information underlying the
specifically
tailored data may need to be aggregated, hidden, removed, or appropriately
filtered to create a
summarized decision model for said decision makers in order to achieve a
rapid, cost-saving,
informed decision. Currently said data is generated by human experts and
manually reduced, or
said information is arbitrarily reduced without a specific process and is
subject to human error. In
cases where sufficient detail is deemed to exist, said detail may also relate
to a decision maker's
conditional acceptance of any given solution, thereby providing means for
negotiation in reducing
CA 02258383 1999-O1-08
-2-
cost of the selection, and means to assess the benefit and challenges
represented by the selected
alternative. Said means, because of the overwhelming complexity of a typical
problem, does not
often provide adequate detail for detailed negotiation between parties. This
can be important to the
eventual implementation project resulting from a decision, an important final
part of an advisory
service.
More specifically, the present invention relates to a novel decision support
process
that determines decision making procedures through a series of appropriate
graphical and verbal
prompts, and provides adequate detailed meta-data without necessarily
revealing other attributes
which may be judged inapplicable, confidential, proprietary or simply
overwhelming in volume and
complexity to a decision-maker. In addition, such a system could in of itself
create objects to enable
a means to evaluate multiple variables within a multicriteria decision
problem, and cause from the
results of the method an action or provide reasons for a selection of at least
one alternative amongst
at least two alternatives, or issue a signal thereto to a human operator or
device for decision and
control action. Such systems are known as automated advisory and control
systems, and constitute a
decision support system and process means for customizing a process and
aggregating,
disaggregating and analyzing data which may result in conditional decision
making. In a control
process, this can provide for conditional systems control.
DESCRIPTION OF THE PRIOR ART
Decision support systems are well known, and many methods for their use are
known. Commonly, in decision support systems information is entered which may
consist of a
plurality of criteria arranged in a hierarchical tree to demonstrate
relationships amongst the criteria,
and one attribute related to each criterion to determine the importance of the
criterion, commonly
called a weight. At least two alternatives which are to be compared and rank
ordered are scored
(sometimes referred to as rated) usually against the lowest level leaf
criteria in the hierarchy. The
ratings and weights are aggregated in a mathematical formula to give a final
score to rank the
alternatives. In addition, other meta-data attributes (which may include text
and links to objects over
a computer network) with comparative qualitative information are added to
assist a human decision
maker in qualifying a decision, or selecting an alternative that may not be
the first rank by the final
CA 02258383 1999-O1-08
-3-
score and may be selected when meta-data is provided that mitigates the score
and causes
preference for another alternative.
Generally, such hierarchical decision models are evaluated on the basis of a
single
mathematical formula, requiring either the decision maker to understand the
formula so presented,
or taking in blind faith that the formula directly represents the aspirations
of the decision maker, and
sum-totals the decision. Often this is not the case and rule-based or other
means are required to
better represent the evaluation of alternatives.
To mitigate some of these problems, rule based expert systems such as that
disclosed
by Rhonda L. Alexander et. al. in US patent 5182793 "Computer-aided decision
making with a
symbolic spreadsheet" have been developed. However, such rule-based systems
are classically
complex, often requiring many rules that can confuse and do not address the
clarity of the decision
process. Obviousness of the process is masked by the complex number of rules,
and rigidity of the
process where such rules are essentially "Yes/No", precluding gray areas of
understanding or
negotiable points.
F. N. Burt in US patent 4829426 "Computer-implemented expert system and method
for decision-making" discloses a method of eliciting qualitative and verbal
information by a mentor
from a decision maker, and through a method of reduction by comparing
alternatives a criterion at a
time, arrives at a conclusion to select one, using Boolean, ordinal, cardinal,
and qualitative
information, from an expert human or data from a storage device and using
equations provided and
elicited by direct involvement of a human mentor (facilitator) to evaluate the
alternatives. Such a
process is well known among professional facilitators and has been criticized
since the influence of
the mentor can result in bias from mentor influence, and can also obscure the
final result from the
decision maker, since the mentor or facilitator must reduce the information
for presentation and
simplification purposes, often forcing "yes/no" responses to many questions
that have more
complexity to them.
CA 02258383 1999-O1-08
-4-
Decision making software such as Expert Choice [manufactured by Expert Choice
Inc.,5001 Baum Blvd., Suite 650, Pittsburg, PA, USA, 15213] and ERGO
[manufactured by
Arlington Software Corporation, 740 Saint Maurice Street, Suite 410, Montreal,
Quebec] also
provide the means to evaluate alternatives to arrive at a decision, including
only text information as
meta-data. However, no means is provided by which the decision maker is
assisted through the
decision process based on prior knowledge of other decision makers, and prior
knowledge can also
provide qualification information for selecting alternatives other than the
top ranked alternatives.
The volume of such information can be very large, and summarized information
is a painstaking
process of peeling down the data, as F. N. Burt demonstrates in US patent
4829426.
A decision support system for analyzing client satisfaction with vendor
choices is
disclosed by Richard H. Case et. al. of Gartner in US patent 5734890. The
patent discloses a means
to aggregate ratings and weights from a number of decision makers using a
plurality of criteria with
the intent on enhancing client satisfaction through a systematic and thorough
methodology for
collecting information and data related to weights and criteria. However, such
methods are based on
spreadsheet techniques, and the volume of data such methods can handle is
limited by practical
considerations. In addition, Gartner Group subsidiary Decision Drivers
Incorporated [56 Top
Gallant Road, P.O. Box 10212, Stamford, Connecticut CT 06904-2212] has
developed a systematic
and automated process based upon Expert Choice decision support software in
which data is
accumulated separately in data bases and spreadsheets, then the aggregate data
is passed to the
Expert Choice decision aid tool by manual cut and paste techniques and tools
available in Microsoft
WindowsTM operating systems. This process can take several hours to days or
even weeks. Further,
not all the information is made available due to space considerations,
technological inability, the
physical impracticality of providing all information through 'cut and paste'
methods, or for
proprietary reasons. The current methods are thus very limited, and data
integrity is easily
compromised leading to potential loss of faith by clients in the efficacy of
the method. Further, a
full analysis can be limited by the size of models as judged from the number
of criteria and
alternatives the software can accommodate. for example models with 16,000
criteria and a dozen
alternatives or more are likely, and may often be composed of component models
which in
themselves can be viewed as complete decision models for specific
applications. Further, there are
significant limits due to numerical computation time in large models, model
size and structure
CA 02258383 1999-O1-08
-5-
limits imposed by the decision support tool and additionally, the problem of
determining all
pairwise comparisons due to its pairwise methodology (a problem well
documented and noted by
many authors - for example, Millet and Harker, "Globally effective questioning
in the Analytic
Hierarchy Process", European Journal of Operational Research, 1990, 48, 88-9
and Olson and
Dorai, "Implementation of the centroid method of Solymosi and Dombi", European
Journal of
Operational Research , 1992, 60, 117-129]. The simplification process causes
the "invisibility" of
all the underlying data as it is removed, and manual summarizing processes are
easily subject to
human error where employed, nor can the models be easily tailored to specific
clients - indeed, it
can become physically and economically impractical to do so with the manual
process. The "data
invisibility" can further make the process appear as a 'black box' process,
and leaves many
questions unanswered. Prior art has therefore given cause to doubt in the
efficacy of the process,
resulting in limited applications and low market penetration of currently
available systems.
Automating this process would therefore provide a great advantage, and enable
automated decision
tools to be more widely accepted, as well as proving the efficacy of the
process. Automating such
decision tools gives a major advantage to such a player in the new industry
called the Automated
Advisory Services Industry.
In Expert Choice a means to compare alternatives exists in which the weighted
differences of value are summed to produce a single number designating the
advantage of one
alternative over another from a selected parent node. Such a method is also
described by by Richard
H. Case et. al. of Gartner in US patent 5734890 in which the summed
differences between products
is called the 'Competitive Edge' of one alternative over a less valued
alternative. This is an obvious
process and is used in such processes as providing final quality rating of one
alternative compared
to another that is assembled from comparing component criteria. However, the
competitive edge
calculation is unrelated to the worth and cannot, for example, provide a
direct costed negotiating
point with a vendor in order to obtain a best price. This is a point
completely missed by Case et. al.
Also, other comparisons are unavailable, and only two alternatives at a time
are compared, limiting
the use of the process. Further, many decisions are not 'unidimensional' and
require examination of
worth in respect to target alternatives such as industry standards, and ideal
alternatives; such means
is not described in the method. Additionally, simple differences are
inadequate to describe more
complex decision strategies available, for example, in Arlington Software's
ERGO 4.0 product.
CA 02258383 1999-O1-08
-6-
(ERGO 4.0 has no aggregate equivalent method for analyzing worth by summed
differences across
alternatives: it can use a more complex methodology based on pattern analysis
as well as several
scenario and sensitivity schema such as the addition and removal of criteria
based on weight, which
gives rise to decision strategies beyond simple direct cost and benefit
comparisons). A more
sophisticated methodology is required to translate differences that are not
only linear summed
differences in alternatives against a simple linear weighted sum decision
strategy, but against a
decision strategy obtained from a cluster of appropriate questions, and
represented by a set of easily
interpreted charts and numbers by one familiar with the art. Said decision
strategy may be
performed by those less knowledgeable in the art, and yet who may be the major
decision makers.
I O Providing assistance on an as-needed just-in-time (ANJIT) basis can be
provided with through
networked technologies. Providing said ANJIT services would give significant
advantage to a
company providing services in the automated advisory services industry,
particularly if said
services are integrated with an integrated approach to the decision support
system and the generated
model. Further, a model containing sufficient but not overwhelming information
that may
I 5 dynamically provide adequately accurate scenarios by providing a client
with simple procedural
interfaces would enable different strategic decision scenarios to be examined.
Such a tool would
need to integrate decision support and decision making tools such as ERGO and
Expert Choice with
optimization and data compression methods to achieve rapid and sufficiently
accurate results.
Currently, no such system exists.
Since the selection among alternatives can require thousands of criteria in
any given
model, aggregating data is a necessity in order to make an informed decision
feasible within a
reasonable timeframe for an expert and non-expert alike. No decision aid is
available which can
adequately cover all known available data and aggregate the information
automatically to enable
thorough decision analysis. Further, no guidance is given in how to build a
decision strategy that
mitigates risk, and often there is a requirement to seek expert assistance in
order to make such
decisions, often adding significantly to the expense of the process as well as
delaying the decision.
It is tacitly assumed the expert has a thorough knowledge of the area, and can
present an unbiased
view. However, it is not possible for a human or even group of humans to
handle the volume of data
available without an automated tool given the complexity of many decisions.
Additionally,
'unbiased' data implies the reconciliation of differing views and experiences
from at least two
CA 02258383 1999-O1-08
_7_
sources of information, set in the context of the current decision, and this
can represent a problem to
said expert in terms of retention of past experience, maintenance of
information concerning the
field, and frequent reconciliation of differing and contradictory information.
Without an appropriate
tool, said expert cannot claim to be unbiased. Often said experts are linked
with one or two
preferred vendors, and have deeper knowledge of said preferred vendors than
other vendors. Thus it
can be said that most experts in the field cannot provide an unbiased view of
all available
information, and have no tool to back up such a claim. As well, aspects of the
detailed analysis may
be proprietary or confidential or not relevant in a specific decision, or
necessary in order to compute
aggregate values, and consequently a means to tailor decision models 'on-the-
fly' may present
I O significant advantage. Tools such as Expert Choice have limited
capabilities in this respect - indeed,
it is not even possible for some attributes - particularly in regard to
hiding/revealing specific data
and. aggregating data in at least one node in the hierarchical tree. ERGO has
several tools which
provide flexibility to the model building process such as Append and Save sub-
trees, but none for
automatically creating aggregate and hidden nodes, with meta-data information
embedded in the
aggregate fields for example.
For automated advisory services and automated decision and control process,
such
knowledge based aggregating processes and knowledge aggregating systems would
add a
significant competitive advantage, providing clients an unbiased and
systematic process for making
decisions in highly complex environments. Currently, no such system is known
to exist.
In addition, systematic and learned methods maintained in a knowledge base may
mean that valuable techniques and methods are passed on to those who may make,
or may assist in
making, similar decisions. Such methods may be scripted, and said scripts
customized to enable
both a client and an expert of an automated service to follow set procedures,
and benefit from a
store of past experience by others. As well, such processes enable the
systematic training of
decision makers and experts more amenable since case studies are well
documented. Controlled
access to said knowledge bases and tailored scripted procedures also provides
additional revenue
leverage for an automated advisory service. Thus an integrated approach to
knowledge base
CA 02258383 1999-O1-08
_g_
generation, script customization, training of clients and experts, centered
around means to make
decisions, would be highly beneficial.
As well, no decision support tool provides a systematic means to aggregate,
accumulate, and store the methods and procedures by which decision makers
reach decisions,
including the practical experience and true costs of, for example, a missing
vendor feature, and the
constraints employed to assist in price negotiations as another example, or
methods by which
specific issues are resolved with particular vendors and clients. In providing
such means to
systematically describe decision processes, and storing said decision
processes in a knowledge base,
and providing a decision maker the means to utilize this knowledge in a
structured fashion, would
result in a significant advantage for any decision maker. It would also apply
to any process making
decisions where consequences can measurably result, and where said processes
can benefit by prior
knowledge of procedures used for making similar decisions. Said process is
also of value to vendors
seeking to improve performance, and provide direction for future product
development in
competitive markets since it would provide the ability of knowing how clients
select their products
and competitive products. The information for the said knowledge base may be
garnered from non-
proprietary and proprietary sources, from clients, public literature, vendors,
and other sources
general by at least one expert in the field.
Said knowledge base may include attributes specific to at least one
alternative, and
may provide at least one significant summary information a client of an
advisory service may find
relevant. Such information as the tradeoff a vendor may have made, knowingly
or unknowingly,
between feature capability and complexity of the product, compared to other
vendors, may be of
significant value in understanding the challenges that may be faced by the
client, and may be of
interest to a vendor who may find at a glance the summarized data and position
of the vendor
product with respect to the industry. Other attributes may be related to costs
and business risk,
vendor market positioning and future vendor development goals, and so forth.
In some cases, some
products may be composed of a cluster of products, and hence a hierarchy of
products, possibly
designated in terms of an alternative hierarchy, may provide a breakdown of a
vendor's offerings in
CA 02258383 1999-O1-08
-9-
component form. To date, no such decision aid capable of examining
hierarchical alternative
structures exists.
In another aspect of said disclosed system and process herein, such data and
meta-
data may consist of engineering information related to risk in a control
process such as that in a
nuclear reactor where many thousands of nodes represent a hierarchical control
structure, the
accumulated risk as components of a system deviate from a norm (constraint)
being so estimated by
the decision support system, and possibly assisted by prior knowledge from
similar situations in
other systems. Data and meta-data for and aggregated node in the hierarchy can
be continuously
updated, the advantage thereby being that operators are not inundated by
information in a large
complex system, and only such data is revealed by a process of node
disaggregation when for
example a local constraint is exceeded or a particular pattern of information
occurs below the node.
Decision support processes for automated advisory and automated decision and
control systems are
of the same process in either case in terms of the required automated
aggregation process. In
1 S automated decision support processes, however, disaggregation may not be
available, or available
by special license, thereby providing added service feature options by the
automated advisory
service. Again in current systems, said service feature options are not known.
Data and meta-data for a knowledge base may be garnered from experience,
statistical reliability data, and so forth. It is also obvious that it would
be advantageous to minimize
the intervention of a human expert, who may have specific biases from limited
knowledge and other
interests, through a systematic process of model customization and use of the
accumulated
systematic knowledge in said knowledge database.
In all current decision support systems text information alone is provided,
and direct
connections "on-the-click of a mouse" to expert opinion provided by means of
direct object linking
to a source of information which may be local or remote to the user, are not
directly incorporated as
part of the decision model, and updating of said information is not available
by any automated
process.
CA 02258383 1999-O1-08
-10_
In another aspect, aggregation of numeric data would mitigate problems related
to
rapid and accurate dynamic feedback to develop what-if scenarios 'on-the-fly'
for large and complex
systems. Said dynamic analysis uses values and parameters of the model - such
as weights and
ratings - which are changed 'on-the-fly' to evaluate the effects on a
particular decision or evaluation
of at least one alternative, usually compared to at least one other
alternative. Because computation
times can require at least several seconds or even minutes for such large
models, such dynamic
analysis is not feasible for interactive use, and is difficult to perform even
with the highest speed
desk-top systems once model sizes and number of data values, plus updating of
chart information
on a graphical display for example, are included. If the information is
aggregated, the computing
problem would significantly be mitigated. For some parametrics such as
Arlington Software's
Matching Index and Composite Index, disclosed in US patent number US5844817
the calculations
require sophisticated aggregation techniques not possible with any current
decision support system.
A means to mitigate the computational time of numerical calculation for large
numbers of criteria
through pre-calculation and storage of aggregated computed parameters at
specific nodes in a given
model would provide a significant advantage in reducing computational effort,
and significantly
extend the use of an aggregate model where data has been aggregated to high
level nodes. In
advising said client about possible scenarios, interactive graphical
procedures would provide rapid
answers not available in any current system.
In another aspect of deficiency, all models created by decision support
systems have
unlimited use in terms of time, and can be distributed without restriction to
any who may have the
appropriate machine readable code to read said model. Protection of
proprietary information is
compromised, and unlicensed use of the model content is therefore at risk.
Currently, only machine
readable code that executes on a processor may be time limited or restricted
to a particular
processor. A means to limit the period of availability of a model, its data
and meta-data content,
features available to a decision maker, and to limit its distribution, would
provide greater control
over the model use, and enhance the revenue generation for an automated
advisory or automated
control system. In addition, prevention of unauthorized input and output of
said models is required,
preferably by encryption and decryption method.
CA 02258383 1999-O1-08
-11-
In addition, a customized set-by-step guideline for making a decision in the
context
of a particular client would provide the said client and advisory service with
a clear procedure and
process. A means to determine a step-by-step procedure, and execute it,
incorporating in at least one
of the means to mitigate said problems described herein, would be highly
advantageous to an
automated advisory service.
To mitigate at least some of the deficiencies in current systems and
processes, a
novel and comprehensive decision support system and process for knowledge
accumulation,
knowledge aggregation and disaggregation, and process for its controlled use
with knowledge based
systems in automated decision processing, is disclosed herein. The decision
support system
provides a means to aggregate data and meta-data, determine decision
procedures and related costs
and risks, and disaggregation techniques. Means to prevent unauthorized use of
model and model
data is further described. As well, a method and process is described for
extending these means and
processes to automated large decision and control systems, and process means
to provide a
customized step-by-step customized and packaged procedure.
SUMMARY OF THE INVENTION
It is an object of the present invention to provide a novel decision support
system
and method which obviates or mitigates at least some of the above-mentioned
disadvantages of the
prior art. It is a further object of the present invention to provide a novel
article of manufacture that
obviates or mitigates at least one of the above-mentioned disadvantages of the
prior art, and provide
process means to take advantage of the present invention.
According to one aspect of the present invention, there is provided a computer-
implemented decision support machine for comparing two alternatives,
comprising:
memory means for storing at least one decision data structure with each said
decision
structure having a plurality of decision criteria organized in parent category
nodes with child
(descendant) nodes;
CA 02258383 1999-O1-08
-12-
input means to assign to each said decision node a weight value to represent
the
importance of each said node to the decision, the plurality of weights and
nodes comprising a
predetermined two-dimensional benchmark pattern;
input means for inputting a first plurality of scores for a first competing
alternative
to the decision factors of said decision data structure, and input means for
inputting a second
plurality of scores for a second competing alternative to the decision factors
of said decision data
structure;
input means for entering meta-data for at least one node, wherein said meta-
data
consists of text information and can further include means for multimedia
links to local objects and
can also contain links to remote objects and experts located on a computer
network or
telecommunications network;
input means for entering at least one raw score which can be ordinal or
cardinal data;
input means for determining how at least one raw score is translated into a
utility
value representing the worth of the raw score toward said decision for at
least one said alternative;
processing means to compute the utility value from the raw score value for at
least
one of said alternatives;
input means for identifying at least one node as an aggregate node into which
descendant criteria attributes below the aggregate node are aggregated into
attributes in the
aggregate node;
input means for identifying at least one node as an aggregate hidden node into
which
descendant criteria attributes below the aggregate node are aggregated into
attributes in the
aggregate hidden node;
input means to determine an aggregate hidden node and all of its descendants
is not
to used in determining the decision outcome;
input means to expose a aggregate hidden node including all its descendants in
the
hierarchical tree, causing said hidden nodes to be used in determining a
decision outcome;
input means to determine at least one descendant node below an aggregate
parent or
aggregate hidden parent node is included in the aggregation to the parent
aggregate node;
CA 02258383 1999-O1-08
-13-
input means to determine at least one meta-data attribute to be aggregated for
at least
one descendant node;
input means to determine the aggregate meta-data field where at least one meta-
data
from at least one descendant node is aggregated;
output means to indicate the status of a node and distinguish by means of a
visual
marker on a graphical display the determined aggregation status of the node;
processing means to aggregate at least one attribute of at least one
descendant node
in the aggregate node in accordance with the aggregate node status;
input means to assign at least one code determining the limitations on access
rights
to said aggregate model, said availability being determined by an authorized
site;
input means to assign at least one unique identification code to said
aggregate model;
input means to remove at least one node and node attributes from said
aggregate
model;
input means to determine at least one attribute fir at least one alternative;
input means to cause said at least one alternative attribute to be retained in
the model
following aggregation;
input means to cause at least one alternative attribute to be dependent on the
value of
at least one other model attribute;
processing means to determine at least one parameter that determines at least
one
aggregate score for an aggregate parent node;
processing means for transforming (i) the first plurality of scores into a
final
aggregate score including using at least one aggregate and aggregate hidden
node score parameter
and (ii) the second plurality of scores into a final aggregate score including
using at least one
aggregate and aggregate hidden node score parameter;
input means to cause said decision support system to select from the available
alternatives a 'best set' of alternatives determined to best meet client
requirements;
input means to determine an industry standard alternative, said alternative
having a
plurality of scores determined from the plurality of scores of all scored
alternatives;
CA 02258383 1999-O1-08
-14-
input means to determine an alternative-specific standard, said alternative
specific
having a plurality of scores determined from the plurality of scores of at
least two selected
alternatives;
input means to determine at least one report template;
output means to store at least one report template;
input means to input at least one report template;
output means to output at least one output signal corresponding to at least
one of the
first and second final aggregate scores to provide a ranking of said at least
two competing
alternatives.
input means to determine at least one constraint for at least one decision
factor;
processing means to determine comparison between at least one alternative
score
attribute value and at least one constraint;
input means to determine at least one constraint from at least one score of at
least
one alternative;
processing means to select at least one 'worst constraint' condition and at
lesat one
'best constraint' condition where at least one worst constraint and at least
one best constraint may
be computed from input data and from a selected set of alternative data;
storage means to store at least one decision object with at least one
constraint
attribute as a decision scenario object with constraints;
processing means to compare the score of at least one decision factor for at
least one
alternative against at least one constraint;
processing means to determine an aggregate of all values of comparison to
provide at
least one final value representing the overall difference value for at least
two alternatives when
compared to the said at least one constraint;
output means to output at least one output signal corresponding to at least
one
aggregate value differences to provide a conditional comparison value or worth
of said competing
alternatives.
CA 02258383 1999-O1-08
-15-
input means to input at least one code, said code determining license
conditions on
said model;
output means to store model as a data structure on a storage device, said
model being
designated as an aggregate model; and
processing means to encrypt and decrypt said aggregate model.
Preferably, the aggregation procedures are executed by at least one Analyst
who is
knowledgeable in the field of the model and its applications. The at least one
Analyst is a trained
expert the model and determines the aggregate model from a larger model.
Preferably, the aggregate model outcome determination method represents as
closely
as possible the required means by the at least one client. Determination of
said decision outcome by
means of a multiattribute value such as the linear sum of weights and scores
is one preferred
embodiment. A second preferred embodiment is the use of pattern match
algorithms such as
Arlington Software Corporation's Percent Match pattern matching algorithm.
Supporting meta-data
for a decision outcome, in one preferred embodiment, is customized to meet
client requirements.
In one preferred embodiment, said at least one aggregation parameter are
obtained
by machine readable executable code on the general purpose processor on which
the model
aggregation procedure is executed. In one preferred embodiment, said
aggregation parameters
enable aggregate values to closely estimate exact values without requiring the
larger data set to
obtain the exact values, said estimation being necessary in the event of
changes of weights, scores
and other values during investigation of various scenarios using said
aggregated model by said at
least one client. In another preferred embodiment, the Analyst provides the
data of the customized
Analyst model prior to aggregation of the said at least one determined
aggregate node, and uses
machine readable executable code on at least one processor such as a neural
net, thereby generating
said at least one aggregation parameter. Said at least one aggregation
parameter is then stored with
the said aggregate model as at least one attribute of the said at least one
aggregate node.
CA 02258383 1999-O1-08
-16-
In another preferred embodiment, the Analyst determines that the aggregation
parameters are not required, and does not generate said aggregation
parameters.
In one preferred implementation, the aggregate data and meta-data is
selectively
output to a printer or disk file. In another embodiment, said aggregate model
is encapsulated in an
electronic message to generate a request for proposal (RFP) to elicit
responses from the 'best set' of
alternative vendors. In another embodiment, said vendors are replaced by
alternative strategies,
actions, project alternatives, and personnel, and other such choices as would
occur to, and be
evaluated by, the at least one Analyst.
In another preferred embodiment, the aggregate data and meta-data is
selectively
output to a printer. In another embodiment said selected data and meta-data is
saved as a disk file
such as a spreadsheet or as a word processor document as is commonly known,
and other known
disk file formats as is available at the time. Said selected data and meta-
data in one embodiment is
transmitted over a computer network. The determined format, however, contains
a summarized
report on a limited number of preferred vendors most likely to satisfy the
requirements of a client.
Said preferred vendors can, in another preferred embodiment, be selected by
automatic process
means, said selection being performed by said decision support system. In
another embodiment,
said preferred vendors are manually selected from the list in the full Analyst
model by the said
Analyst prior to storing the aggregate model. A report on said selected vendor
can in one
embodiment include exception reporting and summarized vendor strengths and
weaknesses, and
other data and meta-data that is deemed relevant by one knowledgeable in the
art.
In another aspect of the current invention, the at least one Analyst
determines at least
one attribute to assign to at least one alternative, said attribute having at
least one value
summarizing at least one aspect of the alternative. At least two attributes
can be arranged in a
hierarchical tree indicating hereditary relationships amongst the said
attributes. In one embodiment,
said hierarchical arrangement determines sub-alternatives. Said sub-
alternatives are scored as
independent alternatives, and said scores are used to determine the outcome
score of the parent
alternative. In another embodiment, said attributes represent a hierarchical
relationship of attributes
CA 02258383 1999-O1-08
-17-
for a single alternative. In either case, said at least one Analyst indicates
by input means said means
by which at least two attributes are to be aggregated and retained in the
aggregate model. In another
embodiment, said at least one Analyst causes said at least one attribute for
at least one alternative to
be provided in the form of a chart. In a further embodiment, said chart is
given in the form of meta-
data, thereby providing a client with the summarized values and meta-data
information applicable to
the specific requirements of the client.
In another aspect of the current invention, the said limits on access and
distribution
at a client site for a provided model may include, but not exclusively, timed
availability, availability
of different features of the model and the decision support system, model
storage location
limitations, limitation of access by specific at least one individual, and
number of licensees with
access to the model in a file server environment, and limitation on the number
of scenarios a client
is allowed to generate. In one preferred implementation the first reading from
storage of said model
by a decision support system supplied to a client causes said decision support
system to prompt for
a second licensing key code to initiate time and feature availability. An
authorizing site may then be
requested to provide the second key code, and may be provided on determining
said client is
authorized. On inclusion of the second key code, the rights and privileges of
access and distribution
to said aggregate mode are accorded for the said provided aggregated model. In
a preferred
embodiment, means to protect from the unauthorized determination of the
internal data structure of
said model is provided by means of encryption and decryption methods
incorporated in said
decision support system.
In another aspect of the current invention, the decision support system is
provided to
the client. In one embodiment of said client provided decision support system,
at least one feature
is removed. In one preferred embodiment, the said removed feature is the
ability to determine
aggregate nodes. In another preferred embodiment, said decision support system
model access code
limitation inputloutput features are removed. In another embodiment, said
client provided decision
support system includes means to determine access rights and limitations to at
least one said
customized model.
CA 02258383 1999-O1-08
-18-
In another embodiment, the aggregate model is provided with a code to prevent
further aggregation by said decision support system.
In another preferred aspect of the current invention, the said provided model
license
may be transferred by a process of license transfer from one processor to
another, thereby
maintaining the number of licenses to a specified number.
In yet another preferred aspect of the current invention, said license may
require
renewal from time-to-time. In this process a validation of the expiration
status of the said provided
model is determined by the client provided decision support system. If the
expiration status
indicates the said provided model has exceeded its limited period of use,
processor implemented
code in the said decision support system may initiate a request for license
renewal, said process in
one embodiment being by automated means over a computer network such as the
Internet. Another
embodiment of said process is an appropriate indication to the licensee of the
expiration status of
the said provided model. In another embodiment, such expiration warnings may
appear prior to
complete expiration as a warning to the licensee, thereby providing the
licensee the opportunity to
ensure continued use of the model.
In another preferred embodiment, the at least one conditional difference value
to
compare at least two alternatives is determined by relating at least one cost
attribute to at least one
criterion node score difference value, said cost usually being expressed, but
not exclusively, in
terms of a monetary unit. In another preferred embodiment, said conditional
value difference may
be in terms of a project success risk value, and in another embodiment said
success risk value is
expressed in terms of a success probability.
According to another aspect of the present invention, there is provided a
computer-
implemented decision support system for comparing two alternatives,
comprising:
memory means for storing at least one decision model data structure having a
plurality of decision factors organized in parent category nodes with child
(descendant) nodes,
CA 02258383 1999-O1-08
-19-
input means to assign at least one weight to at least one decision factor
input means to input at least one alternative;
input means to input at least one plurality of scores, said at least one score
being
assigned to at least one decision criterion for at least one alternative
processing means to determine at least one multiattribute score for at least
one
alternative from said at least one plurality of scores for said alternative;
input means to input meta-data into said model;
input means to determine at least one chart comprising at least one data value
in said
model;
input means to determine at least one report template;
input means to assign at least one meta-data attribute to said chart;
input means to input meta-data to at least one meta-data attribute field,
where said
meta-data includes at least one instruction on using said chart, thereby
causing said chart to be
determined as a decision object;
input means to input meta-data in at least one other met-data attribute field
of said
chart;
input means to assign said at least one decision object to a visual decision
dictionary,
said visual decision dictionary comprising of at least one decision object;
input means to input meta-data with at least one instruction on the use of
said visual
decision dictionary;
input means to determine at least one visual decision dictionary with said
input meta-
data as a decision procedure;
input means combine at least two decision procedures, thereby determining a
decision process, wherein said at least two decision procedures are executed
sequentially as in an at
least two step process;
input means to input meta-data into decsion process wherein said meta-data
provides
guidance on said at least two steps;
CA 02258383 1999-O1-08
-20-
input means to remove at least one decision object from said model;
input means to delete at least one decision procedure from a decision process;
output means to output at least one decision object in a data structure;
output means to output at least one visual decision dictionary in a data
structure;
output means to output at least one decision procedure in a data structure;
input means to input at least one decision object into said model;
input means to input at least visual decision dictionary into said model;
input means to input at least one decision procedure into said model;
output means to output at least one output signal representing said at least
one
I O assigned code enabling access to external meta-data transmitted over a
computer network;
input means to input at least one signal representing a request for at least
one
decision object;
input means to input at least one signal representing at least one decision
object;
output means to output at least one signal representing at least one decision
object.
input means to assign at least one decision object, visual decision
dictionary, and
decision procedure to at least one said report template;
output means to output at least one report template as a report data
structure; and
input means to input at least one report data structure.
Preferably said assignment of meta-data instructions and associated meta-data
for
charts, decision objects, visual decision dictionaries and decision procedures
is performed by an
Analyst who is an Expert in the decision area to which the model is applied.
In one preferred
embodiment, said definition and assignment is performed when aggregating the
complete model,
and in one embodiment is saved in the aggregate model. In a second embodiment,
the Analyst edits
the aggregate model to add, modify and delete decision objects, visual
decision dictionaries and
decision procedures. In another preferred embodiment, said at least one
decision object, visual
decision dictionary and decision procedure, report template, customized report
is transmitted by the
CA 02258383 1999-O1-08
-21-
decision support system over a computer network, and are sent from one
decision support system
operating on one processor on the computer network, and read by a second
processor on said same
computer network executing a second decision support system.
In one embodiment of said process, it is contemplated that a decision
procedure,
visual decision dictionary and decision object are identical. In the preferred
embodiment, said
decision procedure contains a plurality of visual decision dictionaries with
meta-data guidance to
guide in the use combined use of said plurality of visual decision
dictionaries, wherein each said
visual decision dictionary consists of a plurality of decision objects, said
visual decision dictionary
containing meta-data instructions concerning use of said plurality of decision
objects in
combination. Said meta-data guidance can be considered in this embodiment as a
set of scripts.
According to another aspect of the present invention, there is provided a
computer-
implemented decision support system for comparing two alternatives comprising:
storage means for an aggregate model data structure;
output means to output meta-data guidance in selecting said at least one
decision
procedure;
output means to output meta-data guidance in selecting said at least one
visual
decision dictionary;
output means to output meta-data guidance in selecting said at least one
decision
object;
input means to select at least one decision procedure;
input means to select at least one visual decision dictionary;
input means to select at least one decision object;
input means to input meta-data in said at least one decision procedure;
input means to input meta-data in said at least one decision object, at least
one visual
dictionary, at least one decision procedure;
CA 02258383 1999-O1-08
-22-
storage means to store at least one decision object;
input means to select at least one decision object; and
output means to store said aggregate model.
Preferably, in one implementation of the process, an Analyst who is an Expert
in the
field of the model determines at least one decision object, at least one
visual decision dictionary and
at least one decision procedure for the decision making process based on prior
knowledge acquired
by facilitation with said client and from experience of the Expert. Said
customized aggregate model
then incorporates client requirements and said client constraints.
In another embodiment, said client further customizes the aggregate model as
determined here. Said at least one Analyst in one embodiment assists in said
customization process.
Preferably, aggregate score and value differences represent as closely as
possible the
aspirations of the decision-maker. Aggregation by linear sums of weights and
ratings is one
preferred embodiment. A second preferred embodiment is the use of pattern
match algorithms such
as Arlington Software Corporation's Percent Match and Composite Index pattern
matching
algorithms.
In another preferred embodiment, said at least two decision procedures are
bound
together to form a complete guide to at least one decision process. Said guide
may be determined by
at least one knowledgeable in the art of making said decision. Input means may
be provided by said
decision support system to build said guide where and said guide may include
at least one of, but
not exclusively,.report template, request for proposal template, decision
procedure, visual decision
dictionary, decision procedure library containing at least two decision
procedures, bid letter
template, and may include meta-data and support meta-data indicating the use
of each decision
procedure constituting said guide.
CA 02258383 1999-O1-08
-23-
In one preferred embodiment, said guide consists of machine readable code
executed
on a special purpose or general purpose processor. In another preferred
embodiment, said guide
resides on a computer network and is accessed by means of at least one output
signal from a device
such as an intelligent paper clip, said clip being provided with an
appropriate code that can be read
by an infra-red or radio frequency device, and said code containing
information on location of the
said guide on the computer network.
According to another aspect of the present invention, there is provided a
computer-
implemented decision support system for comparing two alternatives,
comprising:
input means to select at least one aggregate model;
output means to store all attribute numeric data set of said aggregate model
in as a
structured data set;
output means to determine a copy of said all attribute data set;
input means to select at least one chart;
input means to select at least one object on said chart, said at least one
chart object
representing at least one numeric attribute data value;
input means where input causes said selected at least one chart object to
change size,
position, shape;
processing means to determine said at least one attribute value resulting from
change of size, position, shape of said at least one chart object;
processing means to determine at least one attribute value at predetermined at
least
one interval, said at least one interval being based on size, shape, position
of said at least one chart
object;
output means to output at least one signal representing said at least one
changed
position, shape and size of the at least one chart object;
output means to transmit at least one signal to at least one other chart in
the current
chart set, causing said chart to determine at least one attribute value
assigned to said at least one
other chart;
CA 02258383 1999-O1-08
-24-
output means to update the display on a graphical display device of said at
least one
chart objects currently displayed;
input means to select one copy of a chart, said copy having the unchanged
attribute
data set determined from stored set of data;
input means to determine said copy of a chart attribute data is not caused to
change
as a result of at least one change made to at least one chart object in at
least one other chart, said
copy of chart is then said to be locked;
processing means to determine at least one value representative of the
difference
between at least one selected locked chart attribute data and at least one
said changed chart data
attribute;
input means for meta-data input to at least one decision procedure;
output means to insert date and time data in said decision procedure to
determine a
history of said changes, said changes being thereby stored for future
reference;
output means to output at least one signal representing the changed data;
temporary storage means to store set of changed attribute data as a structured
data
set;
input means to store at least one set of said changed attribute data; and
storage means to store at least one set of said changed attribute data as a
scenario of
said aggregate model.
Preferably in one implementation of the process, the client uses the method to
devise
and compare different decision scenarios. In one preferred embodiment, the
attribute data and meta-
data are assigned to report templates to determine at least one report, said
at least one report in one
embodiment being output on a device such as a printer. In another embodiment
of said process said
at least one report is stored in a disk file, and said disk file sent as an
encapsulated item in an
electronic message over a computer network..
CA 02258383 1999-O1-08
-25-
In another embodiment, said at least one scenario is stored as a distinct
model in a
data structure, where said model represents one scenario. Said data structure
in another embodiment
enables selective access to at least one attribute datum, thereby providing
means to develop further
scenarios consisting of attribute data from at least two sets of attribute
data. In a preferred
embodiment, said further scenario consists of one set of decision factor
weights, and one set of
alternative scores from at least one other scenario.
According to another aspect of the present invention, there is provided an
automated
advisory service process comprising:
means to store data in a knowledge base, said knowledge base being a
structured
data set determined for a specific type of decision and consisting of at least
one decision object;
stored data structure consisting of at least one non-aggregate model, said
model
consisting of a plurality of decision factors arrange in a hierarchical
relationship, a plurality of
alternatives, a plurality of alternative attributes, a plurality of reconciled
scores where reconciled
scores are determined from scores from at least two sources, for each of said
plurality of
alternatives, and a plurality of meta-data indicating at least said contents
of said model;
means to determine industry standard attribute values from said at least one
non-
agregate model;
means to determine the performance of individual alternatives;
means to determine client requirements;
means to provide client with at least one decision procedure;
means to provide client with at least one decision process;
means to determine at least one best alternative to meet said client
requirements;
means to customize said at least one non-aggregate model;
means to provide alternative overviews to assist client in a learning process
in said
decision;
means to license said aggregate model for said client, wherein said license
imposes
limits of access on said aggregate model;
CA 02258383 1999-O1-08
-26-
means to input at least one decision procedure in said aggregate model;
means to customize reports for said client
means to assist said client in report generation;
means to assist said client in generating a request for proposal;
means to assist said client in management review processes;
means to assist said client in letter of bid;
means to assist said client in negotiation procedures and tactics;
means to assess said selected alternative in terms of alternative performance;
means to provide continuous update to said knowledge base and said non-
aggregate
model; and
means to distribute said updates to subscription clients.
In the preferred embodiment said at least one Analyst uses said decision
support
system to aggregate and customize an aggregate model for said client. Said
means to provide
services in the preferred embodiment uses at all stages said decision support
system. In another
embodiment, a separate knowledge base is used to store decision objects,
constraint data, meta-data
and other data as may occur to one knowledgeable in the field, said data being
determined from
prior experience of said at least one alternative. In the preferred
embodiment, said alternatives are
vendors. In this embodiment characterization of said vendors and said
characterization of Client
provides means to compare future clients against vendors, and can be used to
determine expected
success attributes and attributes wherein challenges are likely to be faced by
said client on the
selection of at least one specific vendor.
According to another aspect of the present invention, there is provided a
decision
support system for use with a controlled process or system, comprising:
storage means for a decision data structure having a plurality of decision
criteria
organized in parent category nodes with child (descendant) nodes;
CA 02258383 1999-O1-08
-27-
means to assign each said decision node a weight value representing its
importance
to the decision, the plurality of weights and nodes comprising a predetermined
two-dimensional
benchmark pattern;
means to assign a first plurality of scores for a first competing alternative
to the
decision factors of said decision data structure, and means to assign a second
plurality of scores for
a second competing alternative to the decision factors of said decision data
structure and;
means to assign reference meta-data for each node;
means to assign a method to translate at least one score into a utility value
representing the worth of the said at least one score toward said decision for
at least one alternative;
processing means to compute at least one utility value from at least one score
value
for at least one of said alternatives;
means to determine at least one node as an aggregate node;
means to determine at least one node as an hidden aggregate node;
means to determine at least one hidden aggregate node as not to be included in
determining at least one decision multiattribute value for at least one
alternative;
means to expose at least one hidden aggregate node including all its
descendants;
means to determine at least one descendant node below said aggregate parent
node
as a node whose at least one attribute is aggregated in the parent aggregate
node;
means determine at least one meta-data attribute to be aggregated;
means to assign at least one meta-data field determined to be aggregated to at
least
one meta-data aggregate field in the aggregate parent node;
processing means to aggregate attributes of selected descendant nodes.;
storage means to store aggregate model as a structured data structure;
processing means to determine at least one parameter that determines at least
one
aggregate score for said at least one aggregate node;
processing means for transforming (i) the first plurality of scores including
said
determined scores determined from at least one aggregate node parameter to a
multiattribute value
CA 02258383 1999-O1-08
-28-
for the first alternative and (ii) the second plurality of scores including
said determined scores
determined from at least one aggregate node parameter multiattribute value for
the second
alternative;
output means to output at least one output signal corresponding to at least
one of the
first and second multiattribute scores to provide a comparison of said
competing alternatives;
means to determine at least one decision procedure;
means to store at least one decision procedure;
means to select one decision object;
means to input at least one constraint for at least one aggregate model
attribute;
means to determine at least one dependent relationship between said at least
one
constraint and at least one aggregate model attribute;
means to determine said at least one constraint from at least one attribute
value of at
least one alternative;
means to determine at least one 'worst constraint' condition and at least one
'best
constraint' condition where worst and best constraints represent extreme
acceptance conditions;
processing means to compare at least one score for at least one alternative to
at least
one constraint condition and output at least one value for the comparison;
processing means to determine an aggregate of all said values of comparisons
to
provide a final value representing the overall difference value of at least
one alternative when
compared to the at least one constraint;
output means to output at least one output signal corresponding to at least
one of the
value differences;
input means wherein input of at least one signal causes said at least one
hidden
aggregate node to expose said at least one hidden descendant node;
input means wherein input of at least one signal causes said at least one
hidden node
descendant nodes to be hidden;
CA 02258383 1999-O1-08
-29-
input means wherein input at least one signal causes generation of at least
one
decision factor, said at least one decision factor representing an input
element in said control system
for said decision support system; and
input means wherein at least one signal causes the removal from said model of
at
least one decision factor, wherein said decision factor represents said output
source of said input
signal.
In one preferred embodiment of the process, at least one Analyst or persons
knowledgeable in the field determines at least one decision procedure and at
least one constraint for
at least one decision object. Said at least one procedure and at least one
constraint for the control
process is based on prior knowledge which is acquired from facilitation with a
client, and is
additionally determined from experience with same or similar automated
processes. In another
embodiment an automated system provides as output for at least one human
operator at least one
decision procedure wherein attribute values and meta-data represent the
aggregate information and
the decision making precepts and status of the automated system. In yet
another embodiment, at
least one exception datum is output when an attribute value exceeds said at
least one constraint. In
one embodiment said exception output is to a graphical display device. In
another embodiment said
output signal is directed to an automated process or system. In a further
embodiment, said at least
one exception is output to a printer, and can further be output to a disk
file, and further, said disk
file can be encapsulated in an electronic mail message, said message being
transmitted over a
computer network.
BRIEF DESCRIPTION OF THE DRAWINGS
Embodiments of the present invention will be described, by way of example
only,
with reference to the accompanying drawings, in which:
Figure 1 shows a block diagram of computing hardware embodying a preferred
embodiment of the present invention;
Figure 2 shows a pictorial representation of an embodiment of the computing
hardware of Figure 1 indicating a hardware automated decision and control
process;
CA 02258383 1999-O1-08
-30-
Figure 3 shows a pictorial representation of a second embodiment of the
computing
hardware of Figure 1, indicating computer network client/server capability and
input/output
devices;
Figure 4 indicates the aggregation and tailoring procedure of reducing a large
model
to a client specific aggregate model with aggregate and hidden nodes;
Figure 5 shows the procedure used to construct decision objects with meta-data
and
numeric data;
Figure 6 shows the procedure to compute conditional values such as costs from
constraint information specific to client and to the alternatives;
Figure 7 shows an example of the different fields and visual cues for the
aggregate
status of criteria and preview of meta-data aggregated text fields;
Figure 8 illustrates means to preview aggregate text fields, and graphic
images that
may be saved as objects in the meta-data;
Figure 9 shows means to assign aggregate status for the node;
Figure 10 shows the means to customise the aggregation process by selecting
the
different fields to aggregate, what is to be aggregated, and where the
aggregate information is to go.
It also indicates how the aggregated evaluation is to be represented and
evaluated;
Figure 11 shows the way in which different chart objects may be dynamically
changed in order to produce dynamic evaluations and scenarios;
Figure 12 shows a decision procedure in which rating values are adjusted, and
how a
chart value may be locked, refreshed and in order to control the comparison;
Figure 13 shows the report generation stages of the procedures to generate
RFP's for
example, and final reports;
Figure 14 shows the licensing and license management process at a client site,
and at
an authorisation site;
Figure 15 shows the license transfer process to transfer a license from one
processor
to another.
CA 02258383 1999-O1-08
-31-
Figure 16 shows one part of a visual decision dictionary / decision procedure
composed of decision objects and means by which criteria attributes may be
used in chart data;
Figure 17 shows a graph of weighted average value against the absolute sum
difference function, indicating the approximation method results from the
optimised selection of a
fitting function;
Figure 18 shows a visual decision dictionary in which chart objects are
divided
between data modification charts and impact charts, and where meta-data
technology is used to
provide assistance and guidance to the client;
Figure 19 shows an alternative hierarchy and associated attributes showing how
attributes are inherited for a composite alternative;
Figure 20 shows means to assign attributes and attribute values and properties
such
as equations relating to other attribute values and attributes of factors in
the model tree;
Figure 21 shows the attribute summary table with attribute values for the at
least one
alternative, and means to filter alternatives by said attributes;
Figure 22 shows an overview of a scripted process for assisting a client
through a
procurement selection process, and means by which continuous improvement and
updating of
model information is obtained;
Figure 23 shows the part 1 of the procedure for initial client needs
specification and
procedure for tailoring model and decision procedure scripts;
Figure 24 shows part 2 of the procedure for further detailed analysis of
vendors with
assistance from at least one Analyst;
Figure 25 shows parts 3 and 4 of the procedure wherein vendor selection is
finalised,
bid letters are provided to vendors, and proposal-counter-proposal procedures
may be used. Part 4
indicates means for post-vendor selection and follow-up on execution of
project in order to provide
not only added revenue, but means to update information in the area of the
decision model;
Figure 26 shows means a decision process guide by which a decision support
system
may provide a guide through one or more steps in a decision process
CA 02258383 1999-O1-08
-32-
Figure 27 shows a decision procedure governing an automated process in which
decision objects may be employed to reveal or hide data regarding the
condition of a system.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
Figure 1 shows a block diagram of a decision support system 20 in accordance
with
a preferred embodiment of the present invention. System 20 includes processor
means 24, input
means 28, output means 32 and storage means 36.
In a first preferred embodiment, as shown in Figure 2, processor means 24 is
an
embedded control processor, such as a Motorola 68HC 16 and associated
circuitry 40; storage means
36 comprises a ROM memory 44; input means 28 comprises a series of sensors 48;
and output
means 32 comprises a known controller 52 for a process which produces
appropriate output signals
56. In the embodiment illustrated in Figure 2, controller 52 can, for example,
be a rocket vehicle
launch controller and sensors 48 may comprise launch status sensors, type of
rocket (manned,
unmanned) containing meta-data and data input signals, service request
counters, timers, etc. and
radio receivers for remote site data input. This specific implementation of
the embodiment of
Figure 2 is discussed in more detail below. In otherwise similar preferred
embodiments, wherein
system 20 is used to control a number of different activities in an industrial
process, for example the
control of different chemical process reactors, system 20 may be implemented
from appropriate
discrete components. In such an embodiment, input means 28 can comprise one or
more
appropriate sensors 48, for example position, load, demand, velocity, pressure
and/or temperature
sensors, and database information for each of a plurality of reactors, and
output means 32 will
comprise one or more means to generate control signal outputs, for example
variable voltage signals
to control do motor speeds, solenoid valve or brake actuation signals, etc. If
required, the signals
from sensors 48 can be translated into an appropriate format for processing by
processor and
associated circuitry 40 by, for example, an analog to digital converter, a
protocol converter (in the
event that the sensor signals are provided over a network), or other suitable
conversion means as
would occur to those of skill in the art. In addition, output means 32 may
provide at least one input
to a second decision support system wherein the input may control data and
meta-data presentation
3 0 to human operators.
CA 02258383 1999-O1-08
-33-
As used herein, the term process is intended to encompass commercial,
industrial
and other processes and to include, without limitation: mechanical and/or
electromechanical
operations; chemical process control; HVAC (heating ventilation and air
conditioning) systems;
robotic systems; aerospace flight control; biological injection systems;
medical monitoring, control
and/or alarm systems; power station and electrical power delivery and supply
systems; vehicle
systems and transportation control or any system where a pattern of inputs is
analysed and a
decision process is invoked to deliver a pattern of actions and outputs to an
apparatus or system
through mechanical, electrical or other means.
In another preferred embodiment, illustrated in Figure 3, processor means 24
is a
general purpose processor and related circuitry 60, such as an Intel Pentium
family processor;
storage means 36 comprises a mass storage device 64, such as a Winchester-
style disk drive, a
removable media storage device 68, such as a 3.5 inch high density disk drive,
and RAM and/or
ROM memory 72 which is operably connected to processor 60; input means 28
comprises a
keyboard 76 and/or pointing device such as a mouse 80; and output means 32
comprises a video
display terminal 84, such as a SVGA graphic display, and/or a printer 88, such
as a HP LaserJet IV.
In some circumstances, it is contemplated that input means 28 and/or output
means 32 may
comprise a communications link 90, either in addition to or in place of the
above-mentioned
components, and communications link 90 may be in the form of a local or wide
area network, a
radio link, etc. which is also operably connected to processor 60. In the
embodiment illustrated in
Figure 3, system 20 can also embody a graphical user interface, provided by an
operating system
such as Microsoft Windows 95 operating system, executing on processor 60. It
is further
contemplated that multimedia input means comprising camera 91, sound recording
microphone 92,
and output means of graphic images displayed on at least one graphical display
terminal 84 and
sound from at least one speaker 93 connected to said general purpose processor
60, may be used to
record and may be used to play images and sounds accordingly. It is also
contemplated that input
and output means may be provided by transmission signals 94, 97. Said
transmission may consist of
electromagnetic emissions such as visible light, radio frequency and infra-
red, and may consist of
sound waves. Said transmission may be passive from said monitor 84 and may be
converted and
stored into meta-data as provided by an instrument such as a C-PenTM 98. It is
further contemplated
CA 02258383 1999-O1-08
-34-
said input and output may be realised using a video camera 91, or special
purpose input/output
device such as a transceiver 95 and device 96. In one contemplation of 96,
said device may be an
intelligent paper clip providing direction to at least one data source on a
storage device. Said device
96 may also provide data to at least one data file located on a storage
device. It is also contemplated
said data and meta-data input and output means may be used over a computer
network using said
network connection interface 90 for providing for at least one output signal
and at least one input
signal with at least one other node on said computer network.
In either of the above-mentioned preferred embodiments, system 20 receives
various
inputs through input means 28, processor 60 acting on these inputs according
to instructions stored
in storage means 36 and providing one or more outputs, via output means 32,
which outputs
recommend and/or implement a desired selection between alternatives and/or the
operating state of
the process or apparatus under consideration.
The term decision factor is used herein to refer to a factor or criterion used
as an
element to construct a hierarchically related tree. Said tree contains a root
decision factor that is the
base of the tree from which all other decision factors are descendant decision
factors. A parent
decision factor has at least one descendant decision factor below it. A leaf
decision factor has no
descendant decision factor. A decision factor with a parent is called a child
of he parent decision
factor. Decision factors which directly share the same parent are called
herein sibling decision
factors. Said decision factor tree elements may also be referred to as nodes
and as criteria, said
meaning and all respective definitions within this paragraph remaining the
same as decision factor.
The term "aggregate node" is used herein to determine a designated decision
factor
that has at least one descendant decision factor and where at least one
attribute is determined from
at least one directly descendant (child) node attribute to be aggregated in
said aggregate node
attribute. Said descendant decision factors and all said attributes of said
descendant child factors are
removed when an aggregate model is stored or the aggregate node is caused to
aggregate. Said
determined attributes from descendant decision factors are retained in the
selected at least one
parent aggregate node at least one aggregate attribute.
CA 02258383 1999-O1-08
-3 5-
An aggregate parameter as used herein is a parameter used to determine at
least one
attribute or multiattribute value to a determined adequate degree of accuracy
when aggregated data
is removed during an aggregation process. Said at least one aggregation
parameter provides means
of representing lost attribute data, and may be regarded as providing means
for a compression
algorithm representing said missing data. In one embodiment, said at least one
aggregate parameter
may be assigned to an individual aggregate node following aggregation and
removal of descendant
nodes of said aggregate node. In another embodiment, said at least one
aggregation parameter may
be related to at least one alternative-specific attribute. Said aggregation
parameter is useful in
dynamic data changing where said aggregation parameters provide means to
represent missing node
attributes such as ratings and weights of said missing descendant nodes. In
another embodiment,
said aggregation parameter may determine a mathematical formula, set of rules
relating attribute
values, and at least one table of values. In another embodiment, said
aggregation parameters may
represent a set of coefficients to be used in a default mathematical equation
determined as most
likely to accurately represent missing attribute data in an aggregate model.
As used herein, a hidden aggregate node is a node that is present and may be
displayed on an output device such a graphical display, and may provide
visibility for selected data
and meta-data from at least one descendant node, where said at least one
descendant node is hidden
(i.e. is not exposed and is invisible) on any display device such as a
graphical display. Attempting
to expand the tree below the hidden aggregate node cannot expose said
descendant hidden nodes.
The descendant nodes are physically present in the aggregate model. The hidden
node may appear
like an aggregate node and have meta-data fields containing aggregate data. A
hidden descendant
node may be used for purposes of accurate numeric computations and may be
revealed to
selectively show data and meta-data when at least one specific event occurs to
expose the node.
Said event may be the result of at least one manually inserted key, and may be
by means of at least
one manually input value through a graphical user interface, and may be from
at least one input
signal over a network. Said hidden descendant node may contain proprietary
data and may further
contain secret data where it is desired to have said data available for
purposes of determining a
value. By this means said proprietary and secret information is prevented from
disclosure, while use
is made of said data in value aggregation.
CA 02258383 1999-O1-08
-36-
A trimmed node is a parent node from which all descendant decision factors
have
been removed. A typical trimmed node becomes a leaf node on the tree of the
output aggregate
model whose aggregate rating for at least one alternative is determined from
the normalised weights
S and ratings of descendant decision factors that are removed on trimming. In
one embodiment said
aggregate rating is the normalised weighted average determined from descendant
node ratings and
normalised weights. No other information may be presented or retained in a
trimmed node.
A rating or score as used herein is a value assigned to a decision factor to
represent
the worth, impact, direct value, or relative amount of the factor for at least
one alternative toward
the decision. Said value may be a cardinal or an ordinal value, or a verbal
expression identified with
at least one numeric value. Generally, a rating is a value applied to leaf
decision factor, and a score
is the value determined from combining weight and rating value of at least one
decision factor. In
another embodiment, a rating value may be translated into a utility value by
means of a utility
function or rating method. A 'standard' score or rating may be an industry
standard score or rating
obtained from published industry standard target and requirement data. In
another embodiment, said
standard may be determined by the decision support system from data among all
alternatives rated
in the model. In another embodiment a sub-set of alternatives may be selected
to provide a
'decision-specific standard'. If the alternatives are vendors, said sub-set
standard values are referred
to as vendor-specific standards. In one embodiment, said standard may be an
average value. In
another embodiment, said standard may be determined as for example a
statistical mean or mode.
The terms utility method and rating method as used herein define a
computational
means of transforming at least one raw input value for at least one criterion
to at least one utility
value, thereby providing a sense of worth of the at least one raw input value
toward the goal of the
current model on the basis of a common scale of utility. Said raw rating is
applied for at least one
alternative for which said criterion is applicable. The said raw rating value
and utility value may
further be translated into a verbal statement to indicate in more familiar
terms the worthiness of the
assigned value toward the model goal.
CA 02258383 1999-O1-08
-37-
As used herein, the term alternative represents a physical item, or may mean a
choice
between at least two actions to perform. Said definition is intended to
include the case wherein no
physical alternative or action is selected, and may represent for example at
least one state of a
system against which a performance measurement is made, and said selection of
an alternative
represents a means to measure a system's performance for said alternative.
In one preferred embodiment, said alternative may represent a vendor product,
and
said vendor product may be considered composed of a set of at least two
vendors component
products, said at least two vendor component products being thereby evaluated
individually and
said vendor component products may then be combined accordingly to provide a
composite
solution, and said composite solution may incorporate at least one other
vendor product or
component product. In another embodiment, said alternative may be at least one
system state, said
at least one system state may be composed of a set of at least two sub-system
states. The
performance of a set of inputs may then be rated for each said system state to
determine by numeric
means the validity of said system state and said at least one sub-system
state.
The term decision factor attribute is used herein to designate a value or meta-
data
field that is directly related to at least one decision factor. An alternative-
specific attribute is an
attribute that is directly related to at least one alternative. In one
embodiment said alternative
specific attribute may be determined from a rating or score value of at least
one decision factor in
the hierarchical model tree. In another embodiment, said rating and weight
attribute of a decision
factor for at least one alternative may be determined from at least one
alternative-specific attribute.
As used herein, a constraint is a value assigned to node to which at least one
node
attribute for at least one alternative is compared, the outcome of which
generates a numeric output.
In one embodiment said output may be in the form of meta-data which in of
itself may be useful
and may provide information to assist in determining the suitability of an
alternative. In one
preferred embodiment, said output may not of itself eliminate an alternative
in a selection process,
and is therefore a weak constraint in this sense. In another preferred
embodiment, said output may
CA 02258383 1999-O1-08
-38-
become a strong constraint wherein in of itself said output may eliminate an
alternative from the
decision process. A typical strong constraint may be a mandatory requirement.
Meta-data as used herein refers to non-numeric information that is required to
qualify rating, weight or other numeric data, and may provide other non-
quantitative information
useful in determining the suitability of an alternative in the context of a
decision, and may contain
instructive information on the usefulness of the model, and may further
contain explanatory text,
sound, graphic image and animated and moving images concerning a decision
factor and attributes
such as weight for the decision factor. Meta-data may also be applied to
result information,
alternative-specific attributes, charts and decision procedures, importlexport
procedures of data, and
report templates and report generation. Meta-data may contain embedded links
to other documents
consisting of text and multimedia reference objects such as sound images and
animation and movie
files that may be located locally and at locations on a computer network, and
may contain direct
links to expert advisors over a telecommunications network which may be by
means of electronic
mail and may include direct interactive voice and image connections such as
those provided by
products like Microsoft NetMeetingTM and WhiteboardTM[ESR1] and may use direct
voice and
images through at least one microphone 92 and at least one local camera 91,
and at least one
speaker 93 connected to at least one general purpose processor 60.
An Analyst as used herein may be at least one expert knowledgeable in the art
of
building decision models, and may be knowledgeable in the areas of application
of the model. A
client is defined herein as one who may use an aggregate decision model in
order to reach a
decision, and may take input from the aggregate decision model to reach a
decision independent, of
or with assistance from one knowledgeable in making decisions in the field to
which the aggregate
model applies. Said client may be a decision-maker, and may be knowledgeable
in the application
area of the model.
As used herein, the term "to tailor" means the act of reducing and customizing
the
data set of a model through a process including node aggregation, elimination
of criteria and
alternatives and said attributes of said criteria and alternatives, causing to
store the reduced model
CA 02258383 1999-O1-08
-39-
as a customised aggregate model. One preferred embodiment of tailoring is by
intervention of one
knowledgeable in the art of tailoring a model for specific applications.
Another preferred
embodiment is the process of reduction and aggregation by means of at least
one signal representing
an automated message which may cause at least one node to aggregate selected
descendant data in
the said model, and may cause at least one node to hide descendant children
and their attributes, and
at least one second signal that may cause at least one node to disaggregate,
and may cause at least
one hidden node to be made visible. Said at least one signal may also cause at
least one attribute of
at least one criterion to change, said attribute being, but not limited to,
the weight of the criterion,
score for an alternative, the indicated relevance of the criterion in the
model, and the presence or
absence, inclusion or exclusion of said criterion in a decision and control
process.
As used herein, a decision model is comprised of a plurality of decision
factors
organized in a hierarchical fashion, and containing information including
values and other decision
factor related data therein, and a plurality of alternatives with associated
data including rating
values assigned to decision factors, weights and meta-data, model goal
information and other
related meta-data as deemed necessary to assist in making a decision by one
knowledgeable in the
art.
A chart is defined herein as a means of representing numeric data and meta-
data in
the form of a table or a graph or combination of both thereof. Typically, a
chart contains numeric
and text data of at least two attributes, said attributes consisting of at
least one numeric value and
may contain at least one meta-data datum attribute contained in the model.
Said chart may further
comprise of a table of rating and score values expressed as verbal or numeric
data.
A chart object as used herein may be a bar or point or other symbol or shape
representing at least one data value.
As used herein, the term 'dragging' refers to the action of focusing on at
least one
chart object presented on a graphic display with an input pointing device such
as a mouse, selecting
CA 02258383 1999-O1-08
-40-
said at least one chart object and causing said at least one chart object to
change in size, position,
shape by moving said pointer. Generally, such dragging may cause on a time
delayed basis re-
determination of the at least one value represented by the at least one new
size, position, shape of
said at least chart one object. "Simultaneous dragging" refers to the act of
selecting at least two
chart objects such as bars or points representing at least one data value in
the chart, and changing
each independently prior to re-determination of at least one value represented
by the said at least
one change in position, size, shape of said simultaneously dragged objects.
Such simultaneous
dragging selections and changes may be performed by providing sufficient time
to select a
subsequent object prior to re-determination of said at least one value. In
another embodiment, said
chart objects may be pre-selected as a group and caused to change by dragging
simultaneously.
Multiple points on a chart may therefore be changed in one dragging operation.
As used herein a decision object may consist of at least one meta-data field
containing at least one instruction on the use of said decision object, and
may include at least one
chart and at least one meta-data instruction on the use of said chart. In one
aspect said decision
object provides output means to instruct a client in the method of proceeding
through a step in the
decision process using said representation provided by a chart or at least one
numeric value or meta-
data value. In another aspect, the said decision object may include meta-data
identifying purpose of
said at least one chart, and may include at least one instruction on the use
of said at least one chart.
As used herein a visual decision dictionary consists of at least one decision
object
and at least one associated meta-data field wherein is entered at least one
instruction pertaining to
the use of the said visual decision dictionary.
A decision procedure consists of at least one visual decision dictionary and
at at least
one associated meta-data field wherein is entered at least one instruction on
the use of said decision
procedure.
CA 02258383 1999-O1-08
-41-
A decision process consists of at least one decision procedure and at least
one meta-
data field wherein is entered at least one instruction pertaining to the use
of said decision process.
In one embodiment of said process, it is contemplated that a decision
procedure,
visual decision dictionary and decision object are identical. In the preferred
embodiment, said
decision procedure contains a plurality of visual decision dictionaries with
meta-data guidance to
guide in the use combined use of said plurality of visual decision
dictionaries, wherein each said
visual decision dictionary consists of a plurality of decision objects, said
visual decision dictionary
containing at least one meta-data instruction concerning use of said plurality
of decision objects in
I O combination. Said meta-data guidance can be considered in this embodiment
as a set of scripts.
As used herein, a decision process consists of at least two steps taken in
sequence,
the said steps being determined by at least one Analyst. In another embodiment
of said process,
said steps may be determined by at least one other who may be knowledgeable in
the art of making
the said decision. A decision process guide is a guide wherein said decision
steps are provided in an
outline format, said each step being provided with means to execute said step
and complete said
step as best as may be possible, as may be determined by at least one expert
in the step procedure.
Said guide may provide guidance for the use of means to execute and proceed
through said step.
As used herein the terms to store or output in reference to a data structure
consists of
the action of processor means in writing said model consisting of a determined
data structure from
temporary storage medium such as random access memory attached to a computer
processor to a
medium regarded as permanent storage such as a CD-ROM, Winchester style hard
disk,
programmable read only memory (PROM), removable flexible diskette, magnetic
tape and any
other medium deemed by one knowledgeable in the art as a permanent storage for
said data
structure. Said storage process in one preferred embodiment consists of means
to encrypt said data
structure on output by means of an encryption process such as PGP (a public
domain
encryption/decryption procedure) or any such decryption technique as may occur
to one
knowledgeable in the art When said encrypted data structure is input into the
decision support
CA 02258383 1999-O1-08
-42-
system as disclosed herein, the decision support system may use PGP decryption
techniques to
decrypt said model structure.
An Authorising site is an authorized entity able to service license key code
requests
from client sites, and that has access to information to validate the
eligibility of a client to a license.
In one embodiment, said entity may provide service through a phone-in help
desk system. In
another embodiment, the entity may be an automated system resident on a
computer network such
as the Internet. In another embodiment, said service may be on an internal
corporate network. Said
automated system may be able to review client license requests automatically,
and transmit at least
one message with a response in regard to said license request, where said
response may be to
provide said license key code, reject said license key code request, and send
a request for additional
and repeat information.
A Client is an entity serviced by an automated advisory service and may make
use of
an aggregate model. In one preferred embodiment, the entity may be a
corporation, a government
organisation, or at least one individual. In another preferred embodiment, the
entity may be a
machine consisting of machine readable code executed on at least one
processor, said processor
having storage means such as random access memory and read only memory, and at
least one
permanent storage device such as a computer readable disk.
A licensee as used herein refers to a Client who has applied for and been
granted a
valid license by at least one authorising site.
In one preferred embodiment of the invention, a decision support system is a
processor implemented code that, in accordance with Figure 1, the processor 20
reads the code from
a storage device such as a Winchester type disk 64 or from a removable storage
medium such as a
3.5" disk or CD ROM 68. An Analyst using said code causes said processor to
read data
representing at least one decision model from at least one storage device. The
decision model may
have been designed, assembled and populated with data and meta-data by at
least one Analyst
CA 02258383 1999-O1-08
-43-
familiar with creating such models in the field to which said model applies
.In one embodiment,
said populating may be done using said same decision support system. At least
one Analyst
knowledgeable in the art of tailoring said models, and knowing the purpose to
be accomplished by
current said model, proceeds through the process 101 indicated in Figure 4.
The Analyst selects a
node using a pointing device such as the mouse 80 in Figure 3 or by keyboard
entry 76 using arrow
keys and a selection key on the keyboard, and indicates at least one criterion
node at which
aggregation is to take place such as Manufacturing 250 in Figure 7. In the
current preferred
embodiment, the Analyst may select functions from a menu and sub-menus 350 and
351 of Figure 9
actions to designate said selected node as an aggregate node or as a hidden
node, or may trim the
current node, its status so indicated by a mark or symbol 250 beside the
designated node. In another
preferred embodiment icons designating said aggregation functions may be
present in for example,
at least one Microsoft Windows standard toolbar for the model tailoring
window. Descendant nodes
whereof the data is to be aggregated may be indicated by other symbols 251.
The Analyst may
select at least one other node such as 'Supply chain management' 253 wherein
at least one
descendant node and said descendant node's data and meta-data may be hidden
but retained in the
aggregated model. The Analyst may further select for each aggregate node
specific descendant node
data to be aggregated as indicated in figure 10 401 and apply further
conditions to further filter
descendant node attributes to be in or out of the aggregation procedure 404
and 405. Such
conditions may include the descendant factors with the best and worst scores
for at least one
alternative, the descendant factors which most deviate from the average score;
determining the level
below the aggregate node at which descendant factors are represented in the
aggregate node 408. If
descendant node level does not extend to the lowest descendant node, the meta-
data from the lowest
descendant nodes may not be aggregated included in said aggregation node. In
the present preferred
embodiment, only descendant node attributes directly descendant (the level
below) from the
aggregate node are aggregated as a default condition, all descendant data from
the lower nodes
being aggregated transparently (i.e. aggregated and removed on aggregation
without choice of
preservation of specific parameters for calculation and choice of aggregating
meta-data information
other than the, on one preferred embodiment, the weighted average and other
parameters as required
time-to-time by the decision support system). Aggregation may be selected to
go to lower levels
408. The Analyst may select the aggregate node field 403 into which meta-data
and summary
numeric data are placed for client review. The Analyst may select any of the
available data and
CA 02258383 1999-O1-08
-44-
meta-data fields 401 for aggregation into at least one of the aggregate node
fields selected in 403,
and include specific items such as multimedia files and direct ('hot') links
to experts via E-mail or
dial-up line 410, and may include a specific tutorial related to the node. A
restriction of selecting
only one aggregate field for each sub-factor meta-data field may be imposed as
a preferred
embodiment. In addition, the Analyst may select how the scores and weights may
be represented in
the aggregate field by selecting specific rating and weight interpretation
methods 405, 406. For
example, a score of'90' may be translated into the verbal score of'Excellent',
and a score of '30'
may indicate a'Poor' score. A weight which is over 5% may be regarded as 'very
important', while a
weight of 0.2% may be considered 'Unimportant'; the most important factor may
be identified by
the verbal statement 'Most Important', and the least by 'Least Important', and
so on. Said rating and
weight interpretation methods may be created and edited as part of the
customisation process 407.
The verbal scores and weights may be combined into sentences such as indicated
in 256. Numeric
values may also be included 409 to show at least one score and at least one
weight, and at least one
verbal statement to indicate the sense of the value 255. A sentence in a
selected aggregated meta-
data field 403 may then read
"Production Planning is the most important factor [2.1 % of total decision] in
Manufacturing and scored poor [ 30 out of 100] for Barndon Plus Corp.'
Score information may be written, for example, a typical output may read:
'Manufacturing scored Average for alternative SoftWares Inc. [53.8], Good for
Barndon Plus Corp. [68.2] and Poor for Pertall Office Suite [42.4] and carries
a weight of 5.75%
toward the decision. For further source information contact Harry.Bi~ s
a,arlin~~soft.com or at URL
www.manuf.com/~Bi~~s or call 999-123-4567'
'Design and Engineering is the least important factor and scored Excellent for
alternative SoftWares Inc. [93.8], Fair for Barndon Plus Corp. [62.3] and Very
Good for Pertall
Office Suite [82.4] and carnes a weight of 0.75% toward the decision.'
The major three weaknesses of SoftWares are in order of importance:
Master Production Planning - Poor [48.4] moderate weight [0.75%]
Inventory Management - Very Poor [32.3] average weight [0.58%]
Production Configuration Management - Average [51.9] weight [0.55%]
CA 02258383 1999-O1-08
-45-
The major three strengths of SofWares Inc. are in order of importance:
Computer Aided Manufacturing - Very Good [78.9] high weight [0.82%]
Computer Aided Design - Excellent [87.9] moderate weight [0.70%]
Product Formulation - Good [65.4], low weight [0.35%]
For specific information consult document [X]'
[X] may be an embedded icon in the document and whereupon pressing said icon
results in the document being opened by the appropriate application assigned
to the file type such as
a file with extension'.doc' may be opened Microsoft Word. Said document may be
a text word
processing document, and may contain sound and images for further explanation
and additional
contact information. Such information is of use to clients that may need
summary information
prepared in the form of a report 803 in Figure 13 for presentation purposes to
senior officials or
others responsible for such decisions at the client site. In another
embodiment, a Request for
Proposal (RFP) 804 may be generated by the Analyst for the client following
aggregation. Said RFP
may be generated from the aggregate model. Question and Instruction meta-data
fields such as
'Score Question' 300 in the descendant factors and aggregation node meta-data
fields may assist in
generating questionnaires. Said reports may be generated by the decision
support system as a meta-
data output, and may include charts generated by the client. Reports may be
generated and sent to
an output device such as a printer 88, or encapsulated in an electronic
message and sent by
electronic mail over a communications network using a network interface 90.
In some instances, at least one minimum requirement may be established from at
least one constraint. In this case, at least one exception report may be
generated for at least one
alternative in the aggregated data and said at least one exception report
submitted to the at least one
client. Such information may be selected 401and assigned to the selected
aggregate meta-data fields
403, and may be generated automatically by the decision support system for
automatic presentation
to the Analyst. The Analyst may then generate exception reports and may submit
said at least one
report to the at least one client.
CA 02258383 1999-O1-08
-46-
In some circumstances, the Analyst may recognise an alternative may not have
at
least one feature, said at least one feature being represented by at least one
decision factor, and
these at least one feature may be compensated by at least one other related
decision factor, and it
may be desirable to avoid reducing the value of the alternative for a
particular category of criteria
containing the at least one decision factor representing the at least one
missing feature. The Analyst
may then mark the at least one criterion as 'Not Applicable', causing weight
of said at least one
criterion representing said at least one absent feature to be distributed to
sibling and related criteria
in the model parent category that compensate for the absent feature, thereby
mitigating the
undesirable result of reducing the value of the category containing said at
least one missing feature
for said alternative.
On completing the aggregation selection and configuration tasks the Analyst
may
store the aggregate model to a storage medium such as a Winchester drive 64 or
removable medium
such as a CD-ROM or removable disk 68. The model may also be exported to a
spreadsheet or
other formatted document 72, and may be encapsulated and transmitted in an
electronic mail
message.
In one preferred embodiment, said nodes designated as aggregate nodes in
figure 4
104 may be caused to aggregate immediately 121, either before parameters are
determined from
106, or after said parameters are determined from 107.
In the current preferred embodiment, storing the aggregated model 110 causes
said
selected descendant aggregated nodes to aggregate data according to the
selection choices made by
the Analyst prior to assigning a unique model identification code 109, and
storing the aggregate
model122.
In a preferred embodiment, the Analyst may select to cause the aggregation
process
to occur on designating the currently selected node as an aggregate node103,
using default
aggregation settings.
CA 02258383 1999-O1-08
-47-
The Analyst may be unable to store the aggregated model under the same name as
the originally opened model 110, and may be forced to save the aggregate model
at a different
location. Said model may be given a different unique code 109, and said unique
code may provide
indication of the parent model from which the model was tailored.
Aggregation parameters may be required for estimating at least one
multiattribute
value used to assess the rank and value of at least one alternative. Said
aggregation parameters
may be used in certain decision procedures and may be used to generate
different scenarios that
may include changes to at least one weight and at least one score, yet such
weight and score and
other data may be removed through the aggregation process. Said parameters
provide means to
approximate values enabling calculation of at least one required
multiattribute value. A
multiattribute value may be any value incorporating a plurality of other
values and used to
determine the rank ordering of a plurality of alternatives. In the current
preferred embodiment, said
1 S multiattribute values may include standard deviation, Weighted Average,
Matching Index,
Weighted Average Composite Index, and Percent Match as calculated in Arlington
Software's
product ERGO. In another preferred embodiment, other common multiattribute
values may be
calculated as required from time-to-time, and may include cost and risk values
that may be related
to at least one criterion attribute and at least one alternative attribute.
In one preferred embodiment, the executing machine readable code causes said
processor to process said tailored model generated by the Analyst, but not yet
aggregated, and
causes said process to determine and store said aggregate parameters 107 for
at least one alternative.
In a second preferred embodiment, said general purpose processor may not be
adequate, and it may be determined that at least one other processor executing
machine readable
code to read the Analyst tailored and not yet aggregated model data, thereby
causing said other
processor to generate at least one aggregation parameter 114. Said at least
one aggregation
parameter may then be stored with the aggregated model 108.
CA 02258383 1999-O1-08
-48-
In a third embodiment of the aggregation parameter process, the Analyst may
determine that the aggregation parameters may not need to be processed, and
thereby saves the
aggregate model without determining said aggregation parameters 110.
In another aspect of the invention, the said at least one Analyst may
determine
relevant additional alternative-specific attributes (as in figure 19 and 20)
of the Alternatives 1300,
1400 that are separate from decision factor attributes. Said alternative-
specific attributes may be
inherited in a hierarchical manner as indicated by 1302 and 1303 in figure 19.
Attributes may be
assigned a name 1401 and may be assigned meta-data information 1402. Other
data may be given
and may include at least one value 1403 selected from a list of at least two
values 1404. An
indication that the attribute may contain numeric data may be given 1405. In
another embodiment,
said numeric value may be related through an equation 1406 to other
alternative attributes 1410 and
may be related to at least one factor attribute, said at least one factor
being selected from the model
tree 1412. At least one factor attribute 1413 may be selected for inclusion in
the equation 1409.
Various mathematical functions may be selected 1411 for inclusion in the
equation 1409. Said at
least one factor and at least one factor attribute and said alternative and at
least one alternative-
specific attribute may be identified internally in the model data structure
through an identification
code. Machine readable code of the decision support system may then be
executed to read and
interpret said identification codes and mathematical functions, thereby
causing said processor to
determine the value outcome of the alternative attribute. In another
embodiment, said value list
1404 may contain text and meta-data. In yet another embodiment, the default
value may be selected
from the list of values, and assigned to the attribute of that alternative if
no other value is selected
for the attribute of the alternative. In another embodiment the at least one
assigned value and at least
one calculated value may be assigned to all alternatives 1407. Said
alternative attribute may be
processed according to selected means 1414. Said values may be displayed as
summary
information for the alternative 1501, and text and meta-data information about
a highlighted
alternative attribute may be indicated in the same window 1502. Further, said
at least one
alternative attribute may then be used to sort and filter at least two
alternatives according to
mathematical or textual relationships 1504 and rules 1505, said rules
combining said at least two
alternative attributes into a common means to sort and filter at least said
two alternatives.
CA 02258383 1999-O1-08
-49-
In a preferred embodiment, a decision support system is provided to the client
wherein said decision support system may have at least one feature rendered
ineffective. In another
embodiment said feature is removed from said decision support system. Said
decision support
system is referred to as a client decision support system. In one embodiment,
said means to
determine key codes and sub-keys may be removed. In another embodiment, means
to determine
aggregate state of at least one decision factor may be rendered ineffective or
may be removed.
Other feature limitations may be determined according to required product
delivery functionality
and may be determined by said at least one Analyst prior to delivery to the
client. Said client
decision support system may further be enabled to input only one determined at
least one aggregate
customised model In a preferred embodiment, said client decision support
system may be
determined as providing a limited set of features that have been determined as
a common
requirement by most clients, enabling a standard client decision support
system to be provided.
In another preferred embodiment, the Analyst may determine prior to final
storing
and providing the aggregate model to the client, determine a security code key
120 for said model.
Said security code key may include sub-keys to determine additional
restrictions on distribution and
availability of said aggregate model, said conditions and limitations being
determined by said client
decision support system on input of said aggregate model. In one preferred
embodiment said
security code key and sub-keys may determine the number of licensees, and may
cause limits on the
number of simultaneous readable copies that can be distributed at the client
site over a computer
network. In another embodiment, said key codes and sub-keys may limit the
number of concurrent
accesses to a single model, the model being located on a single storage
device. In another
embodiment, an additional sub-key may also be provided to limit the number of
storage devices on
which the model may reside at a client site, thereby allowing multiple
concurrent copies to exist. In
another embodiment, said code keys and sub-keys may enable only named users to
input said
aggregate model. Said key and sub-keys restrictions may be activated following
delivery of said
model, and by an additional process initiated between the client decision
support system at the
client site 900 and an authorising site 920, following initial reading of said
model by the client
decision support system. Once the authorisation codes have been input, the
model may be used
according to the conditions in the license as determined by the code key and
sub-keys.
CA 02258383 1999-O1-08
-50-
Processes of model customization may be by means of at least one facilitation
process with at least one decision-maker 111. In another preferred embodiment
of the process,
specifications 112 from at least one expert in the field may be provided to at
least one Analyst, said
specifications providing the basis for the said at least one Analyst to tailor
the decision model. One
such embodiment of an expert may be a consultant in the service of the client.
In another aspect of the process, the Analyst or one Expert in the field of a
specific
aggregate model may execute the machine readable code and select at least one
aggregate model.
The Analyst or Expert may then select at least one chart from a set of charts
150 (Figure 5) or
determine at least one chart therein. In combination with each chart an
appropriate question may be
formulated to ascertain whether a client may require the chart as ari
attribute of the chart 155, and
explanatory and instructional text, meta-data and links may be entered and
edited 156 to correspond
to the said selected chart, thereby customising the meta-data for the client,
and creating a decision
1 S object for said client. Multimedia embedded documents and animated figures
may be included in
the meta-data makeup 165 to provide additional information and clarification
of the use of the chart.
The meta-data may be used to assist said Analyst and Expert and client so that
said
Analyst/Expert/client may effectively utilize the chart to assist in the
decision. Said meta-data can
also serve in training sales staff and future experts and analysts, and
provide means to effectively
communicate the decision process to clients. Each decision object resulting
from said chart and
meta-data instructions thereby renders at least one complete decision aspect
as relevant as possible
toward the decision for the said client. Sometimes a single decision object is
insufficient in dealing
with all aspects of a decision. A visual decision dictionary may then be used
where said visual
decision dictionary consists of at least one decision object used respond to
one aspect of the
decision 158, and may have its own meta-data associated with it 159. Typically
a visual decision
dictionary may incorporate at least two charts in combination. An example of a
visual decision
dictionary is the sensitivity analysis 1200 whereby the sensitivity of a
particular alternative to its
ratings and weights may demand the simultaneous presence of three charts: one
representing
alternative ratings such as Option Rating Bars 1210 in Figure 18, one
representing the model
criteria weights such as the benchmark chart 1211, and one representing the
final evaluation of the
alternatives such as the score breakdown 1212. Preferred embodiments of said
charts are further
CA 02258383 1999-O1-08
-$1-
illustrated in Figure 16 as charts $O1, $02, and $07 respectively. The
sensitivity visual decision
dictionary may be used to observe the effect of changing the weight and rating
individually or
together, to observe the outcome in terms of rank. An embodiment of said
process is indicated in
figure 11 whereby said at least one rating value may be changed by dragging
606, and at least one
$ weight may be changed by dragging 603, and said effect noted in the score
breakdown graph 609,
representing a summary of the value outcome on said alternatives resulting
from said changes of
ratings and/or weights. Meta-data for the visual decision dictionary may be
used for example to
explain the implications of changing the rating of certain alternatives (for
example, if the alternative
is a vendor, in six months the vendor may have added new features and improved
current features to
its product, and hence the rating improvement impact may be assessed), and
weight change
implications of certain criteria may be explained from expert opinion, thereby
making sensitivity
analysis meaningful and realistic. Another such visual decision dictionary may
be to observe the
change in cost-effectiveness and rank of an alternative as rating and cost
information are changed.
A combination of visual decision dictionaries may be provided to determine a
decision procedure
1$ 167 wherein different aspects of the decision - in this case sensitivity
outcome and cost - may be
examined, and the combination provided with meta-data instructions and
guidance in use of said
combination. In figure 18, for example the decision objects illustrated by
1210-1212 may be
replaced by visual decision dictionary titles, and associated meta-data fields
may then refer to the
decision procedure and included visual decision dictionaries. Decision
procedures may further be
stored in a structured knowledge database 162 for later retrieval 161, 164 and
incorporated into
other customised decision models where they be edited and customised for the
another client.
Additional constraint data 163 may also be included and may be incorporated as
separate object
data sets 163, or may be bound with each decision object 16$ in the knowledge
base. The visual
decision dictionary may be stored with the model 160.
2$
In another aspect of the present invention, each dictionary may be accompanied
by at
least one equation 1$7 determined by the Analyst or one skilled in the art.
Said at least one
equation may use the at least one said numeric datum available in the
accompanying chart and at
least one result of the at least one said equation used to assist in
determining at least one negotiation
point for at least one alternative. Said at least one equation may use at
least one constraint 16$
applicable to the at least one chart composing the decision object.
CA 02258383 1999-O1-08
-52-
In another embodiment of the present invention, the visual decision dictionary
may
be selected from a list of visual decision dictionaries in the knowledge base
161. Said selection may
be assisted by at least one question requiring at least one response by one
skilled in the art in order
to select at least one visual decision dictionary from the knowledge database.
Said visual decision
dictionary knowledge base may also contain at least one constraint datum 164
for at least one chart
contained in a visual decision dictionary. The at least one constraint datum
may be determined from
proprietary or confidential data received through interview processes with
clients and vendors, and
from publicly available material.
In another aspect of the art, the decision support system may provide the
Analyst
means to create a flow-through process for the client, said flow-through
process being indicated by
a set of instructions.
In another aspect of the present invention, an Analyst or one knowledgeable in
the
art may have provided code keys and sub-keys as part of a license condition,
said keys and sub-keys
being stored with the aggregate model. On inputting the aggregate model, said
client decision
support system is caused to determine the license validation state of said
model. If the client
decision support system determines that the license requires a second key to
continue inputting said
model 901, said client decision support system issues a request for a second
key, said request may
be output to a graphical display. In another embodiment, said request may be
transmitted by
electronic mail, or may prompt for personal communication between requesting
site 900 and
authorizing site 920. Upon receipt of the request for authorization 921, the
authorizing site verifies
the request is eligible for license 922. If the requesting site is eligible,
the authorizing site issues an
appropriate at least one license code 925. Alternatively, it may provide a
message indicating the
request for license code is refused 924. Said information may be transmitted
verbally or by an
automated response system such as electronic mail 926, or automated signal
over a computer
network to said client decision support system. In one embodiment of the said
process, receipt of an
electronic message containing said at least one license code causes machine
readable code at the
authorizing site to transmit message to the said machine readable code
currently executing on said
CA 02258383 1999-O1-08
-53-
processor at the client site, the said at least one license code may then be
entered by automatic
process through machine readable code at the client site. In another
embodiment of the process, a
licensee may enter the license code manually. In all cases, on acceptance of
the said at least one
license key, said client decision support system verifies said at least one
second key is a valid key,
prior to continuing input. Said second validation key is then stored 906 where
in one embodiment
said stored license may be with said model. In another embodiment, said
validation key is stored in
a location known to the client decision support system. In either case said
model is licensed for use.
If the at least one license key code is not valid, or no license key code is
provided, the model is not
read and the said model file is closed 907. If the license key code is valid,
the model file data is read
by the client decision support system 908.
In another aspect of the invention, the license period is verified 904 for the
said
model, and if said license period is exceeded, said machine readable code may
prompt the licensee
to update the license 903, whereupon the process of authorization 921 through
926, and validation
905 would again take place. The status of the license may, in another
embodiment, be indicated
prior to license expiration, thereby warning the licensee of the time left
before expiration of said
model usage. In another aspect, said license verification process may
determine that the license is
violated during an attempt to input said model by a client decision support
system. In this case,
input is denied for the at least one client decision support system.
In another aspect of the invention, specific features of the model may be made
available for limited time periods. At least one sub-key may be issued to
activate at least one the
feature of the model. For example, hidden aggregated data may be made
available for a limited
period, or the ability to use at least one report template and at least one
specific visual decision
dictionary. At least one sub-key may be issued to de-activate said features,
and may be transmitted
automatically over a computer network. Control of distribution sensitive
information can be
provided through this methodology by at least one central administrative
center with means to
provide said service. Such control may be provided as a service in the
automated advisory services
and decision and control products.
CA 02258383 1999-O1-08
-54-
In another aspect of the invention, machine readable code may be executed to
transfer the model license as in figure 15 from the current processor on which
the license resides
1001, to another processor 1007. The process may involve the following steps:
1. Transfer out of the current model license to an intermediate storage medium
1002 such as a disk
file location on a network, or removable medium: the model may then be copied
from the
current processor, and the license code for the model marked invalid 1005. In
another
embodiment of the process, the model file may be deleted from the current
processor 1006. In
another embodiment of the process, only the license information may be
transferred, the act of
transferring said license to the intermediate medium then being followed by
invalidation of the
model on the first processor 1005. In yet another embodiment, the license may
be transferred
directly to the second machine over a computer network 1003, and said model
may remain
resident on a network server with a shared storage device 1020.
2. Transfer in of the license for the model to the second machine, wherein
said license of the
model may be copied. In one embodiment of the process the license may be
copied into a copy
of the said same model on the second machine which is not currently valid,
said model having
the name and content identical to the model on the first machine. In another
embodiment, the
said model file is directly copied to the new machine, and license information
is modified
appropriately on the said second machine to reflect the validity of the
license on the said second
machine.
3. The limiting on the number of licenses issued to a client is maintained by
the transfer-out-
transfer-in process by adding a restriction. In multiple use licenses, only
the authorized
administrator can initially provide licenses up to the limit set by the
license agreement. In this
procedure, a set number of license keys are provided and given one at a time
by the
administrator up to the limit allowed by the license. From then on, only this
number of licenses
can be transferred through the transfer in-transfer out process 1000.
CA 02258383 1999-O1-08
-55-
4. Further restrictions may be applied to prevent the transfer process to
unauthorized processors by
requesting prior to model release the identification numbers of the computer
processor whereon
at least one client decision support is executed. Thence such transfers can be
limited to specific
machines, said processor identification number being encrypted into the said
model file. In a
second preferred embodiment, the processor identification number is read
automatically on first
installation of the model by the client decision support system executed on
the specific
processor authorized by the administrator, and may not be read by another
client decision
support system executing on a processor with an identification number not
corresponding to the
authorized processor identification number.
In another aspect of the process, the said client decision support system
consisting of
machine readable code is executed at a client site, causing said code to read
at least one aggregate
model. In one aspect of the current invention, the licensee may change at
least one weight, and may
change at least one rating, and may change at least one meta-data field. The
change of at least one
datum may cause said at least one processor to determine a new outcome, and
output said data to a
display device such as graphical display monitor. In so doing, aggregate
values are utilised in the
process.
In another aspect of the invention, a licensee may be provided with at least
one
Decision Procedure in the form of at least one decision object contained in at
least one Visual
Decision Dictionary (VDD) that may be contained in a decision procedure 181 by
the said at least
one Analyst with said aggregate model 182. Through a process of question and
answer, which may
include guided instruction in text and other meta-data formats within the said
Analyst provided
decision procedure, as well as by direct links provided in the help system to
at least one Analyst,
said links consisting of electronic mail services and may further include
interactive assistance using
such methods provided by for example Microsoft NetMeetingTM over a computer
network or
similar, the licensee may select at least one decision object in the decision
procedure and display the
at least one decision object on a graphical display. Said client may then
proceed to analyse the
information contained in the model with said guidance of the meta-data, and
may add additional
information such as constraint data and meta-data comments and embedded data
links. In another
CA 02258383 1999-O1-08
-56-
embodiment of the invention, the licensee may select at least one visual
decision dictionary by
pressing, for example, a tool button 600 in the window provided. Data from the
visual decision
dictionary may then be included in at least one report template, and the
licensee may then select at
least one alternative amongst the at least two alternatives for further
examination and report
generation.
In another aspect of the invention, separate at least one constraint
information may
be determined to characterise the requirements of a client. Said constraint
information may be
entered manually by means of a keyboard 76, thereby modifying or adding at
least one constraint
data item in the said provided model (183 and 185). In another embodiment,
constraint information
may come from a client database, data warehouse or other source available to
the client 184.
Constraint data from all sources may then be combined 187. Said constraint
information may be, for
example, the requirement that a product feature includes the ability to read
files of a specific format.
Said feature may be present to a limited extent -e.g. the file read supports
only a version of the file
format that is lower than that of the version used at the client site, thereby
reducing the
attractiveness of said item, and may cause a need for in-house customisation,
and said customisation
may cause increased cost, and said cost resulting in a negotiation requirement
with the vendor (the
alternative). In this way, such formulae as entered into the decision model by
the Analyst in step
157 that may relate the rating value of the alternative to added cost can be
customised by the
Analyst for the client, and evaluated and noted as a negotiation point in
steps 182 through 190.
In another aspect of the invention, at least one decision object may be used
to
provide information of total cost of ownership. As indicated previously, if a
value does not meet a
specific constraint condition, cost of ownership may increase. In Figure 16,
for example, chart 504
represents a quadrant analysis graph with weight of at least one decision
factor plotted on the y-
axis, and rating value of said at least one decision factor on the x-axis.
Decision factor values on
the left of the cross-hairs 506 in figure 16 are important decision factors
judged as poor performers,
where poor performance indicates a need for additional customisation costs of
a vendor's selection.
The related costs of the poor performance may be aggregated into a net cost,
and added to the total
cost of the alternative. Such costs may then be used for negotiation purposes.
The formulae to
CA 02258383 1999-O1-08
-57_
determine said costs may be embedded with the attributes of the decision
factor as part of the model
data structure. Said formulae and associated values may be inherited by
charts, or assigned and used
in estimating costs and other values in a chart, as executed in step 157. Said
assignment may be
done by the Analyst, and may later be modified by the client using the client
decision support
system. Cost data may be determined from other clients, and vendor projects,
and may be
determined from similar experience at the client site.
In another aspect of the invention, the licensee may determine the need to
operate at
least one decision object in a dynamic mode to explore various scenarios of
attribute values. In the
current preferred embodiment, when the licensee selects this mode of
operation, the data for said at
least decision object is transferred to temporary storage. Said temporary
storage may be a computer
storage device such as a disk data structure, or it may be in the form of
random access memory. In
either case, the data is maintained separate and apart from that in the said
provided model data. In
dynamic mode, at least one data point in at least one decision object chart
representing at least one
variable in the decision may be changed by use of a pointing device and
dragging at least one chart
object to a new position on the chart, or entering new values in a table.
Figure 11 demonstrates at least one example of said dynamic mode. In figure
11 rating values for a plurality of alternatives and decision factors are
represented by the chart bar
objects 606, and said chart bars may be adjusted 610, and the effect observed
as a change in
position indicating a change in the rating, and said change may be transferred
to another chart such
as the quadrant chart where a new value is determined by a change in position
608. Similarly, the
weight and rating of at least one decision factor may be simultaneously
changed, and the change
reflected in the length of the y-value bars 607 and 608. Changing both weights
and ratings of a
decision factor can be accomplished by moving points in the quadrant chart
608, thereby
simultaneously changing the ratings of at least one alternative, and the
weight of at least one
decision factor. Other parameters such as costs may changed as in the
Cost/Weighted Average chart
605 by means of moving the data point chart objects. The calculation update
may be performed'on-
the-fly'. In another embodiment, the updates may be selectively applied to
each chart. Said changes
may then result in at least one what-if scenario to be established. Each new
scenario may then be
CA 02258383 1999-O1-08
-5 8-
saved using a toolbar tool 602, and retrieved using a second toolbar button
601. In another
embodiment, menu items may be added in for example the File menu item, as is
standard practice in
Microsoft Windows operating system applications. In a third embodiment, said
scenarios may be
stored in the same model structure, and presented by means of tabs, and
special charts wherein said
scenarios may be displayed at the same time, and may be compared. Said
comparison process may
follow a scripted procedure as determined by the Analyst or one knowledgeable
in the art, and may
provide grounds for negotiation input by, for example, increasing the rating
of one feature such as
Product Functionality to indicate at least one improvement of the vendor
product feature set. Said
required improvement may then lead to specific analysis of product features
that are lacking, as in
the absence of support for a particular operating system. Said vendor may then
be required to
reduce the price, or provide at least limited support for the operating
system, the details of which
may be provided by the at least one Analyst using the Analyst's more detailed
model and
information content. Said information content may for example indicate the
cost to the vendor in
upgrading said product to provide the support, and may include the anticipated
cost to the client due
to its absence. Either way, said point is available for negotiation, and
negotiations may thereby be
assisted by the at least one Analyst.
In another embodiment, the value outcome of the said changes in ratings and
weights
may be detailed in table format, as indicated in 611. In this embodiment,
values may be entered into
the cells and may then change said values as though the points had been moved
on the chart, and
may cause said chart objects to change accordingly.
In another aspect to the invention, in order to determine the composite index,
a
proprietary algorithm developed by Arlington Software Corporation for its
product ERGO, a special
data aggregation technique is required. One component of the equation relies
on the generation of
summed absolute differences between the expected (desired) contribution of a
criterion to the
decision, and the actual contribution the item provided for each alternative
score, for each
alternative. The aggregate node may contain a linear sum of prior values, but
in the case of the
Composite Index, the final aggregation value cannot be known if any weight or
rating is changed in
the system. This is because the contribution of an aggregated criterion is
based on the original
CA 02258383 1999-O1-08
-59-
weighted average of the original aggregated model supplied by the Analyst, not
the new weighted
average value arising from changes in the data. In order to understand this,
we consider the
determination of the Composite index.
The weighted average is a linear transform and it is easily aggregated. The
Composite Index is different. For aggregation of the descendant data of node L
from descendant
nodes J for an alternative T we may write the composite index in the form
'S jT
w~T h V T ~ S _T
1- +~wir
21-min wn,;n;jT~dl y VT
CI T = 1 + ~ w~T J ...................................................... 1
vj
where w are the global weights normalized to node J, SST is the rating for
node J and
alternative T, and vT is the weighted average for alternative T which is the
sum of the product of
each weight and rating for each descendant leaf node and wherein the leaf node
weights are
normalized and add to 100% at the root node, that is, for NjT leaf criteria
the weights wjT for
alternative T are:
N.n .
wjT =1
...............................................................................
........................ 2
j=1
In the composite index equation, the summations can be divided between the
contributions from the descendant nodes under the aggregate node and all other
nodes. Equation 1
can then be rewritten in the form of the following parameters
ecr + ex
1- +hLT+hX
CIT = 21- min wm",:cr ~ wmin;X ..............................................
3
1+g~T+gX
where the items with subscripts LT represent value contributions from
descendant
decision factors under an aggregate node L, and those with the subscript X are
the value
contributions from all other parts of the model besides L. The components for
LT are determined
by:
CA 02258383 1999-O1-08
-60-
V T = ~ w~T S~T
...............................................................................
........
dJ
2 SLT .. S
hLT = ~ ~w~. V ~
...............................................................................
......
'dJcL T
wmin;LT = min ~wm;~;~T ; b'j ~
............................................................................ 6
gLT= ~w~
...............................................................................
..................~
b'JcL
S eLT = ~ w~. h - S'!z'
...............................................................................
........ 8
JcL V T
where we consider all determined J descendant nodes to be aggregated as being
under parent aggregate node L.
If a weight or rating elsewhere in the model is changed, said change is
reflected in
the weighted average of equation 4. It is seen that equations S and 8 only
depend on the weighted
average. In the case of S, said transformation for the contribution is easily
obtained. If v~T is the new
weighted average, and if the former weight of node L was WLT and the new
weight is W LT then
equation S transforms as:
z
h1 LT - ~ ~ W LT ~ w~T SJT
VJcL wLT v~T
CW LT ~ ~ V T w2 SJT
= Jr _
tlJcL WLT v T ) vT
................................................................ 9
z _
__ w LT _V T ~ 2 SJT
WJT -
WLT ~ VAT VJcLC VT
=~W~LT 1z vT
J - hLT
WLT ~ v~T
In equation 9 it is clear that the values outside the summation are all known.
If hLT
1 S on the right of the equation is retained as a parameter, then it is a
simple matter to determine h LT.
Similarly, the two other parameters in equations 6 and 7 can be preserved as
aggregate node
parameters and transform as simply as in equation 9. However, equation 8
presents a problem due
to the absolute value calculation in the summation. It can be shown that an
approximation to
calculate a LT is given by
CA 02258383 1999-O1-08
-61-
Nc
a LT ~ ~ aLT~JT ~~LT I + eLT ~
.................................................................. 10
JeL
where
V JT V JT
aJT - N _ -
_ jI LT
.......................................................................... 1 1
~,WJkYJT
JeL
c,,T. - ~ a~ _ IJ
...............................................................................
................. 12
However, it is determined that this approximation yields generally errors of
order
10% or worse, depending on the distribution of weights and ratings
distributions of the descendant
decision factors. This inaccuracy is unacceptable.
In order to mitigate this problem a number of procedures are possible. In the
present
embodiment, said procedures comprising:
1. Data compression techniques in which the score data of said aggregated
nodes are
summarized in a set of parameters indicating the distribution of said rating
values of the aggregated
node. Such compression techniques seek to solve the following constraint
optimization problem:
Given a discrete distribution of values f(x~ where x; is a variable (a
weighted
score contribution in our current embodiment, but it may also include a weight
or a cost
distribution for example) over the range ~xm;~, xm~J, then find a function
g(f(x~J, such that
for a given value of x; the value z(x~ is given by
CA 02258383 1999-O1-08
-62-
Zlxi~_~an~,'nV lxi
n=0
and given that the exact solution is
y = v~f'~x; ~]
........................................................................ 13
then optimize gn [ f ~x; ~] subject to
minimum~y - z~~
In the current preferred embodiment, the variable x; is the weighted average
VT of an
alternative T. Typically, the range of VT is [0,100]. The function of concern
is the weighted
absolute sum difference of contributions, that is
.f~VT~=~u'jrl- S'
...........................................:...................................
.14
j=1 V jT
The function g"~f(VT)J is generated by evaluating f(VT) at discrete intervals
of VT.
over the range of VT. This provides a spectral graph 1100 as indicated in
figure 17. In one preferred
embodiment the expression for g"~f'(VT)J in calculating z(VT) 1102 may be
gn(f(VT),=expC-t2un(ym~'-VAT
j~.................................................................. 15
max min
and one may provide means through an approximation method on estimating
coefficients a" through optimization and approximation methods as may be
applied and may be
available in the technical literature, and may be specially developed for
certain classes of score
distributions. Said coefficients are then stored as aggregate parameters for
the aggregate node.
In some instances, if said rating value of the aggregate node is changed, said
at least
one rating value of the at least one descendant node must be considered
modified. For a realistic
view of the rating changes for an alternative, where for example the
alternative is a vendor, said
CA 02258383 1999-O1-08
-63-
rating changes may be provided with a rule base to determine which of said
descendant decision
ratings may be changed for said alternative. For example, if a vendor is
focusing on developing
Computer Aided Design 306 in figure 8 rather than supporting Engineering
Change Management
305, and said decision factors are under aggregate node as indicated, an
aggregation rule would
cause an increase in Computer Aided Design 306 rating prior to Engineering
Change Management
305 rating, until said Computer Aided Design 306 rating reaches the maximum
rating for said
nodes, when the rating of Engineering Change Management may then be changed.
The process of
rating change is subject to an optimization process wherein said rules may be
applied and the value
of equation 14 determined for determined values of the rating. Thence a
function of the illustrative
form in figure 17 is generated and a series of parameters for each new rating
value determined . In
this case, several sets of an coefficients may be generated, each set
representing a specific rating,
weight and weighted average scenarios for each alternative. In another
embodiment, said at least
one rating may determine at least one alternative attribute. In said case,
determination of the
alternative attribute value may further be represented as a function of the
score value of the
aggregate node.
In this procedure the rating values and weights no longer appear in g"'f(VT)J,
and the
equation is purely dependent on the value of VT. The at least one set of
coefficients a" 1103 are
stored for at least one alternative, and generally are specific for each
alternative, and assigned to
said aggregate node. It is normally the experience that the number of
coefficients will be much
fewer than storing the number of raw scores and weights, and hence the data
compression is
achieved. Further it is anticipated that calculation times may be greatly
reduced, and further may
significantly reduce demands on system resources, thereby further assisting in
reducing the
calculation time. Further, said compression may mean that smaller machines
than otherwise may be
able to perform calculations with sufficient accuracy without the requirement
of the full data set.
Figure 17 illustrates means of comparison of z(VT) with the exact solution. In
the current
embodiment, f(VT) = e~T.(VT) where L=0 and where 0 may indicate the root of
the tree.
In the current preferred embodiment, the set of coefficients for each
alternative may
be stored as attributes for each aggregation node. The said coefficients may
be generated by the
CA 02258383 1999-O1-08
-64-
Analyst prior to data aggregation and saving the model in at least one of the
steps 106, 107, 114. In
the current preferred embodiment, the requirement for generation of at least
one aggregation
parameter may be determined by the Analyst
The selection and accuracy of the functions g"~f'(VT)J may depend on the
nature of
the distribution function f(VT). Consequently, it is anticipated that at least
one such function may be
required, where the choice of said at least one function may depend on nature
of the function f(VT).
In another preferred embodiment, said parameter VT may represent cost,
benefit, risk
and other values significant to the alternatives, and may further represent
combinations of said other
values. Summary scores may also be displayed in separate tables, as indicated
by 604, and said
summary scores may be dependent upon the identified node in the hierarchical
tree 604.
2. Ignore the pattern of data in the descendant nodes, and use only the
aggregate
scores as if they were trimmed nodes. Parameters to enable determination of
the Weighted Average
Composite Index (as in equations 5-8), standard deviation, and Percent Match
may be calculated,
but cannot be used in dynamic mode. the aggregate node is treated as if it
were a leaf node in
dynamic mode.
3. Providing means to approximate changes in the distribution of contributions
from the aggregate nodes by modifying the transformations required to
aggregate the data. In this
method, said transformations 5-8 may be modified and approximations sought by
means of pre-
changing the complete Analyst model prior to aggregation. Said resulting
output of values is
modeled by means of statistical distributions. However, said method is
dependent on knowing
which decision factors and which decision factor attributes are likely to be
changed. In one
embodiment, a table of values may be stored with the aggregate model as a
result of said pre-
aggregated generated scenarios. In another embodiment, at least one
statistical distribution may be
determined such as a Weibull distribution for the ratings and weight
distributions, score
CA 02258383 1999-O1-08
-65-
distributions and final outcomes. This method is complex and not well
understood because of the
variety of possible distributions, and is not the preferred embodiment.
In all cases wherein aggregate parameters are determined, the impact on
alternative
attributes may further be assessed. Total cost of ownership of a vendor
product, for example, may
depend significantly on the development direction the vendor is taking for its
products. Hence in
addition to aggregate parameters for decision factors, aggregate parameters
for alternative attributes
may need to be determined, as by means of one of the three embodiments herein
disclosed.
As an example of an application of the automated advisory process making use
of
the model building, tailoring and advisory processes disclosed herein, Client
X seeks to purchase an
Enterprise Resource Planning (ERP) suite of products.
Many vendors offer ERP suites, and the complexity of the suites is large
because of
the many facets that must be understood by Client X. For example, said ERP
system must support
interfaces to a wide range of corporate databases, and be linked to the
databases through customized
interfaces. Said ERP must also be web enabled, and be able to provide reports
for web display, for
the sales force, for administrative personnel, and for reports for senior
management. The ERP must
also be capable in providing up-to-the-minute inventory information, maintain
lists of clients and
integrate sales inquiries, ordering systems, supply requests, purchase orders
etc. Further, Help desk
information is to be tied to product types, help desk personnel performance
measures, and
performance of sales personnel.
Many aspects of the ERP system are vendor specific, with some vendors being
strong in some of these areas, and weak in others. Weak areas provide
challenges to Client X, and
may cause additional costs to Client X. Moreover, the implementation process
is complex, and the
ability of each vendor to deliver such an ERP system must be assessed.
Further, the stability and
likely future of said ERP vendor, and the future direction of the ERP vendor
development, need to
be taken into account. Client X may not have the resources or expertise for
said research, or the
CA 02258383 1999-O1-08
-66-
time and reliable, systematic means to do so. An automated advisory service
(AAS) is approached,
said AAS being a company specializing in independent research on ERP systems.
The AAS may
have developed a model and database that may contain significant information
available from
public sources, private client installations for at least two vendors, and at
least one vendor. The at
least two vendors are independently assessed for their financial stability and
likely future by at least
one Analyst familiar with the field. Said at least one Analyst may have
constructed the said
multicriteria model for an ERP procurement decision process using the decision
support system
herein disclosed, said model containing all possible features of the said at
least two vendors' ranges
of product. Said model may have accumulated several thousand said criteria
representing different
aspects of the decision and may further contain privileged information
concerning the vendor, and
may include product features represented by criteria in the model. Ratings for
said criteria and said
vendor products may have been obtained from former client implementations,
independent
evaluations of vendor products, and other sources that may be found and
included. From the said
decision model, the said at least one Analyst may generate questionnaires and
may provide said
questionnaires to Client X that may be distributed by means of electronic
mail, and by any other
means as may be suitable at the time, including but not exclusively web based
forms and automated
response and analysis processes. An example of an automated response may be
the establishment of
a website document with controlled access, and response data processed and set
into a database
from which the information may then extracted and may be processed through
machine readable
code, and may cause said processor to tailor said ERP model. In a second
embodiment, that may be
additional to the first embodiment, from the return of said questionnaires the
at least one Analyst
may then assess the responses and may accordingly use the information gathered
to assist in
tailoring the said model. In another embodiment of the process, the Analyst
may make use of
interactive computer network group processes to interactively determine said
decision model,
thereby characterizing Client X. In the current preferred embodiment, the
preferred interactive
network product is NetMeetingTM manufactured and developed by Microsoft
Corporation, Seattle,
Washington. The Analyst may then interact with at least one member of Client X
over the computer
network, and may provide the use of the model to at least one Client X member
through the
interactive group process, thereby jointly editing the said model, and may
determine the criteria
required for Client X, and may remove those criteria not required by Client X,
and may further
tailor the said model with other information and determine decision procedures
such as sensitivity
CA 02258383 1999-O1-08
-67-
analysis, as required by Client X, and may be available from the said model
and at least one
knowledge base. In another embodiment of the process, the said at least one
Analyst may visit the
client site and through a process of questioning and elicitation from at least
one decision-maker of
Client X and further characterize Client X. By said means Client X is duly
characterized, and the at
least one Analyst may then determine similar past clients who may have similar
characterization
profiles.
As an example of said characterization measure, a measure of said closeness of
Client X to another client may be determined by means of estimating the
Matching Index value (this
value is determined as part of the composite index in equationl) as determined
from the common
criteria count in the two aggregate models, and weights of said common
decision factors as
determined by Client X and that used by said other client of said advisory
services. Other measures
such as a linear regression fit may be applied, said means being determined by
one knowledgeable
in the art. Said characterization then assists the Analysis in determining the
decision objects and
procedures that may be required by Client X and have been stored in the
automated advisory
services knowledge base.
The at least one Analyst may have used the decision support engine to build
and
populate the model, and may at least one part of said model by executing
machine readable code to
translate the format of the stored model data structure stored in another
structure data format into
the format of the said model and may display said model for tailoring
purposes. Said at least one
Analyst may then proceed from the information provided by Client X, to tailor
the model.
Mandatory data may be tailored, and at least one mandatory component of the
model may be
disabled, and may be removed from the model, and others may remain enabled.
For example, if
Client X uses IBM computers but does not use DEC computers, then all relevant
mandatory
information for IBM computers may be enabled, and those for DEC computers may
be disabled,
and may be deleted from the model, or marked Not Applicable, the said at least
one Analyst
determining the course of action to be performed. Said at least one Analyst
may then designate at
least one node for aggregation. For example, it may be agreed by the at least
one Analyst and at
least one member of Client X that the criterion "Product Functionality" may be
aggregated at the
CA 02258383 1999-O1-08
-68-
second level above the root, as the detail below the node is not required at
this time. It may also be
determined by the said at least one Analyst that the information below the
said criterion "Product
Functionality" is important, but it is desirable to remove them for
simplification requirements, and
therefore aggregation parameters will be calculated for the node. On the other
hand, the item
"Corporate Service and Support" is determined asunimportant in terms of the
information content
below the node, possibly because preferred vendors may be closely the same in
this parameter,
thence the said at least one Analyst may determine that there is no
requirement to generate
aggregation parameters for said node, and consequently no aggregation
parameters are generated,
and the node is simply trimmed, retaining only the normalized weighted average
from all
descendant nodes as representing the score for the said node for each
alternative in the model. In
another embodiment, the evaluation for the ability of the vendor to actually
deliver, under the factor
"Corporate Viability", has score distributions for at least some alternatives
that are not amenable to
current methods to calculate aggregation parameters of sufficient accuracy in
the calculations. This
may be indicated by prior comparisons of exact calculations and current
methods as may exist at the
time in calculating the aggregation parameters. The node "Corporate Viability"
is determined by the
at least one Analyst to be marked 'Hidden', preserving said data for
calculation purposes, while
simplifying the data presentation to Client X members, and retaining
confidential any proprietary
information therein.
Decision factor attributes such as weights may be determined from prior
information
as disclosed earlier. Some criteria weights may be reduced to zero as these
criteria are determined
by the said at least one Analyst to not be relevant to the needs of Client X.
Similarly, costs related to
the presence and the absence of a feature may have been obtained from prior
experience of similar
Clients of particular Vendors. The costs of ownership of particular solutions
from each Vendor suite
of ERP products may further have been garnered from other installation
experiences. Said data may
be present in the Analyst model, and may be aggregated as factor attributes
and Alternative
attributes. On the basis of required attributes and aggregated information, as
well as the agreed at
least one decision procedure to be used for the decision, said at least one
decision procedure is
applied and the at least one Analyst may then generate an initial ranking of
the alternatives.
According to one aspect of the invention, the at least one Analyst may
determine a short list of
vendors who best meet the Client requirements. This may include in one
embodiment selecting
CA 02258383 1999-O1-08
-69-
those Vendors that fail the least set number of mandatory requirements, as
well as those that may
have the highest overall scores, according to the chosen decision method that
may include, amongst
others, lowest cost, best score, and best benefit per unit cost or best cost
per unit benefit. Said at
least one Analyst may then add instructions and other meta-data to the
aggregated model, and this
S may include adding, removing and editing decision procedures, factor
attributes and alternative
attributes. Once said short list of preferred vendors is determined and
decision procedures
appropriately tailored, the aggregate model may be saved, and a report
generated for Client X. Said
report may be customized from a template and include aggregated information,
and may be
distributed to at least one member of Client X. Said at least one aggregate
model may further be
provided to Client X. In another embodiment of said process, a report may be
provided to Client X,
where said report may disclose aggregated model criteria and may show criteria
weights, and may
include preferred alternatives and determined alternative attributes, and may
further include meta-
data. Said report may be generated for management level presentation to Client
X.
1 S In another aspect of the invention, Client X may require the generation of
a Request
for Proposal (RFP) to be sent to the preferred vendors. The Analyst may select
at least one report
template within the decision support system to automatically generate the said
RFP, An RFP is
generated by machine readable code that may be a component of the client
decision support system,
causing said processor to process said model andselect data from said model,
and organize and
format said selected data and output said data in the form of an RFP report.
In one embodiment the
output is in the form of a printed report executed on a printer. In another
embodiment said report is
saved in a known application format such as Microsoft MS Word, and may be
stored on a storage
medium such as a CD-ROM, and may be sent by electronic messaging system to at
least one
member of Client X. In another embodiment, said report may contain at least
one decision object.
In another aspect of the invention, the Analyst may assign and store keys and
sub-
keys in the aggregate model in order to limit the distribution and time of
availability of the model,
as may be required according to at least one license agreement with at least
one member of Client
X. The analyst may then provide at least one member of Client X with at least
one aggregate model
and at least one client decision support system.
CA 02258383 1999-O1-08
-70-
Client X upon receipt of the aggregate model, and having obtained the
necessary
authorization keys from an authorized site, may then proceed to further
analyze the vendors using
the aggregate model and dynamic decision procedures to generate at least one
scenario. Said AAS
may assist Client X to generate a short list of vendors for which RFP's may be
generated from at
least one report template and to whom said RFP's may be sent.
In another aspect of the process, when said RFP responses are returned, the
said
responses may be supplied to Client X and may be provided to the said AAS.
Data may also be
provided automatically over a computer network into which vendors have access,
said computer
network thereby providing means to automatically update data contained in the
automated advisory
services knowledge database. In this manner, the AAS benefits by updating its
information base and
content. As a consequence, the AAS may then provide a subscription update
service to its clients,
including Client X. Further, from the scenarios generated by Client X, weight
and criteria attribute
I 5 information may be obtained by the AAS and added to the knowledge database
for future reference.
Additional decision procedures may also be obtained and added to the decision
knowledge base,
said procedures obtained from the Client that may be by means of agreement and
may be garnered
from Client X's staff. In another aspect of the procedure, the AAS may provide
further decision
procedures and data obtained from other sources, thereby updating the client
aggregate model, said
process being made automatic by machine readable code that causes a general
purpose processor to
modify existing model attributes and content in the said tailored Client X
aggregate model.
In another aspect of the process that may depend on the level of service
agreed to
between the AAS and Client X, assistance in bid evaluation by said AAS may be
provided. Said
assistance may be by presence of at least one Analyst at the client site.
Selection of said vendor may
follow proposal and counterproposal steps, and scripted decision procedures
may be made available
to Client X to provide assistance to Client X in selecting at least one
vendor. Said decision
procedures may require determination of scores, cost information, and
challenges of said at least
one selected vendor, and determination of business risk based on prior
performance of the at least
one selected vendor. When at least one preferred vendor is selected, the ASS
may provide
CA 02258383 1999-O1-08
-71-
additional services to provide a detailed summary report generated from a
detailed report template.
Said report may include graphic, numeric and meta-data in a format suitable
for presentation for
senior management summary as well as detailed information for project
management requirements
that may include cost and project success risk estimations.
In a second embodiment, said assistance may be delivered by remote
conferencing as
earlier disclosed using customized group networking products such as
NetMeetingTM. The said at
least one Analyst may assist Client X by using scripted decision support
procedures, said script may
have been determined for Client X prior to delivery of the said aggregate
model and said aggregate
report, and may be incorporated in the knowledge base of said AAS. Additional
information and
decision procedures, including negotiation approaches may be garnered from
Client X against at
least one vendor by the advisory service, enabling the storage of said methods
for future reference
in the knowledge base and within decision procedures. Said information may
then provide the at
least one Analyst with means to determine other scripts for other clients.
Further, in another
embodiment, said issues raised by Client X may provide vendors with
information to determine
reasons for lost and won sales, and thereby the impact of Client X views of
the at least one preferred
vendor's weaknesses and strengths may then allow at least one vendor to plan
future development
and market strategy.
In one embodiment, said script may provide to Client X prior analysis and
procedures to enable negotiations with said at least one preferred vendor. For
example, the lack of a
feature such as the non-support for a particular E-mail system may result in
an incurred cost in
procurement and adaptation to the new E-mail system. Said cost may be assigned
to the vendor as a
cost attribute dependent on the rating value for the support of said E-mail
system provided by the
vendor. The sum of such costs may then be assessed and may constitute at least
a component of a
report on least one preferred vendor. A means to deduct such costs from the
ERP implementation
assists in reduction of final ERP cost implementation and better planning for
said insured costs.
Thus if the ERP system is assessed at $3,500,000, the average cost from prior
experience at other
similar ERP installations due to the lack of support for the E-mail system is
estimated at $200,000.
Hence negotiation may allow reduction of the ERP system cost to $3,300,000, or
assist the vendor
CA 02258383 1999-O1-08
-72-
in determining that the cost of implementing support for the E-mail system
would ultimately pay-
back since a number of significant future sales may be affected by said
feature absence. In another
scenario, said vendor and Client X may determine that only partial support is
required, hence the
rating value may increase and indicate a cost to the Client is minimal if
vendor supplies said partial
support, and vendor and Client X may then share the risk. Statistics of
failure of sales as a result of
missing feature, or profit margin detriment, maintained in said ERP model, may
provide vendors
with adequate information to determine future directions of development.
As well, in another embodiment of the process, said advisory service may
provide
means to select outsourcing consultants through at least one other aggregated
model, and may assist
in selection of said outsourcing agency. In another embodiment said means to
assist Client X in
selecting project personnel may be offered to Client X as a service. In a
third embodiment, said
advisory service may provide key personnel to assist in decision support
purposes, which may
include but not exclusively procurement of required items, project management
decisions and
project, personnel and vendor performance analysis. In one preferred
embodiment, each said
assistance aspect may comprise of additional models, and a set of said models
may then be
described as comprising a model library, said library thereby consisting of at
least two tailored
models covering at least two aspects of a project process. Said library may
then be treated as a
single model in terms of licensing, and thence the licensing procedures
applicable to models may
then be applied to said model library.
In another embodiment, said advisory services may follow a detailed scripted
process as indicated in figures 22, 23,24, 25, and 26 wherein the service may
be divided into four
parts, the first 1601 may be to determine the requirements of Client X. Within
part I, as indicated in
Figure 23, the market overview may be provided to Client X 1701-1703, wherein
a broad view of
the ERP process may be provided, and said major vendors 1704, 1705 and
criteria 1706 are
presented in broad terms to Client X. Said process thereby educates Client X
as to the major issues
and relative market placement of the vendors with respect to Client X's
requirements. In so doing
the advisory service analyst may obtain sufficient detailed information of
Client X's broad
requirements to begin the tailoring process through a needs definition process
1707, 1708 including
CA 02258383 1999-O1-08
-73-
mandatory requirements 1709 and other criteria 1710. Requirements definition
templates that
provide means to further identify the needs of the client may be generated
1707 and distributed to at
least one Client X member. Responses to said document distribution may provide
sufficient
information to eliminate at least one vendor from the list of vendors through
compliance analysis
1716-1718, 1715. Thence a tailored model may then provide a selection overview
1719 and first
management review documents 1720, 1721. In one embodiment, at the stage 1719
Client X may
receive a client decision support system and at least one aggregate model
under a license agreement.
Part II of the process may consist of further detailed examination of the
reduced number of vendors
1602, and may consist of the steps indicated in figure 24 1800. Through this
process, meta-data and
Analyst assistance is provided to Client X to evaluate vendors and further
customize said aggregate
model that may have been provided to at least one member of Client X, and
interactive means
process over a computer network may be employed as indicated herein to
facilitate said detailed
evaluation. Templates and decision processes 1804-1806 using visual decision
dictionaries and
decision procedures 1807-1811 may assist in further elimination of at least
one vendor, and may
provide a short list of vendors for which a further report and supporting
documents may be provided
to Client X. In another embodiment, Client X at least one member may also
generate said
supporting documents and reports from customized template reports, and may be
assisted by at least
one Analyst 1813-1815. Finally, decision justification is provided that may
include the strengths
and challenges in the selection of said at least one preferred vendor. Various
scenarios may be
explored, said scenarios being performed by use of visual decision
dictionaries and decision
procedures 1812-1818, and may be assisted by the Analyst. Decision
justification documents may
then be prepared for senior management and senior decision makers, 1819-1821.
In the third part of
the said process, the final selected at least one vendor may be provided a
Letter of Bid that may be
generated from templates within the decision support system using selected
data from the tailored
aggregate model, and may be prepared by the at least one Analyst who may
assist at least one
member of Client X . Negotiation procedures with final at least one vendor may
then proceed 1901-
1903, utilizing said procedures and information contained in the said tailored
decision model, and
may further be assisted from meta-data procedures known by the at least one
Analyst, and may be
contained within the knowledge base of said main ERP decision model 1904-1906.
The Analyst
may assist in the final selection through facilitation procedures between the
at least one vendor and
Client X, or through remote conferencing and multimedia interaction as
disclosed herein, and as
CA 02258383 1999-O1-08
-74-
may be available at the time 1907, 1908, and occur to one knowledgeable in the
art. Final
management reports may be generated from templates concerning justification of
the decision, and
may contain details of decision issues, negotiation points and results and
such items as may be
required by senior management 1909-1911. In the fourth and final part of the
advisory service
process 1604, said analyst may assist project planning by identifying the
detailed challenges that
need to be overcome, and identifying deficiencies and strengths within vendor
features that may or
may not be compatible with Client X systems and processes. In so doing, said
Analyst further
obtains information and data for model improvement. Follow-up procedures may
also be performed
to assist Client X in measuring the performance of said vendor, and thereby
this process may
provide added information on cost of implementation, vendor performance and
project risk at the
detailed level, thence giving the opportunity for using said information 1605
in refining the
knowledge base and Analyst model 1606 of the automated advisory service. Said
refinement may
be determined by one knowledgeable in the art and may include industry
standards 1607where said
industry standards may include industry averages and industry means, industry-
accepted levels and
legal requirements imposed on the industry. Said refinement may further
include vendor
performance standard values for at least one vendor and other vendor specific
information 1608,
and client characteristics 1608. Vendor information may be of great interest
to prospective clients of
the vendors as well as other competing vendors, and conversely, client
characterization may be
important for vendors. Thence advisory services may be realized through the
process of systematic
knowledge acquisition and utilization in assisting clients and vendors based
on actual case studies
and scenarios. Such scenarios may thereby be characterized by the aggregate
model containing at
least one decision factor and at least one decision procedure and at least one
decision procedure
attribute that was implemented, and said characterization may provide means of
comparison
between prospective clients of vendors, and of vendor performance in said
client scenarios, and
among all clients of the automated advisory service.
In one embodiment of the decision support system, means may be provided to the
said at least one Analyst to customize all scripts into a single decision
process herein called a
Decision Process Guide, as may be indicated in figure 26. At least one of the
steps as may be
determined by the Analyst may be provided 1951, and selected decision
procedures as may be
determined by the Analyst included 1953, with customized text and other meta-
data 1954 that may
CA 02258383 1999-O1-08
-75-
assist in linking each step and sub-step in the process, and provide direct
access to at least one
Analyst or other information related to the decision process. Said combination
of indicated steps,
decision procedures, and link data and meta-data may be combined to comprise a
customized
decision process guide for the Client. Client X, for example, may require at
least three steps, and
thereby be presented on a graphical display an image as indicated in figure
26. Said image presents
the major steps 1958, 1963 that may be selected in one preferred embodiment by
means of buttons
pressed using a pointing device such as a computer mouse. On selecting a sub-
step which in the
instance of the figure is Market Overview, said vendors and vendor products
may be presented in a
tree 1965, and details of at least one selected vendor displayed elsewhere as
in, for example, frames
1962, 1969. Said views may have been determined by the Analyst prior to
delivery of customized
decision process. Link information in one embodiment may be displayed 1966,
and contact means
added for additional sites information that may include vendor web locations
1967 and expert
contacts 1968. NetMeetingTM interactive procedures may be invoked as may be
desired and
arranged with the Analyst 1960. In the current preferred embodiment, said
Decision Process Guide
may be an independent computer readable executable code and may be supplied to
the client, the
execution of which may then read at least one selected aggregate model at the
client site. In another
embodiment, said tree may consist of the aggregate model decision factors, as
may be determined
by one knowledgeable in the art, and as may be relevant to the data displayed
in the at least one
chart in frame 1962. In another preferred embodiment, the Analyst may
determine certain visibility
features for the said frames, causing meta-data to appear in for example the
chart frame 1962, and
vice-versa. The analyst may, in another preferred embodiment, add or remove
frames from said
decision process guide, as may be determined for each process sub-step of the
Decision Process.
In another embodiment, said Decision Process Guide may be integrated in with
the
decision Support system, and may only be licensed to read selected licensed
aggregate models, as
may be determined by the Analyst, and said licensing procedures as indicated
in 900 may then be
applied to the Decision Process Guide with the additional restriction of
reading only specific
aggregate models.
CA 02258383 1999-O1-08
-76-
In one embodiment, it may not be desired to provide means for said Client X to
further aggregate said model. Thence input means may be provided to the said
at least one Analyst
to disable said aggregation ability of the decision support system. Said means
may be by inputting a
code into the aggregate model, causing said decision support system to prevent
further aggregation
of said aggregate model. Said disabling may be preferred since the aggregation
process may require
external means to compute said aggregation parameters, and said means may not
be available to
said client, thereby compromising the accuracy of the aggregate model further
since aggregation of
an aggregated node into a higher node may cause loss of said parameters.
In another embodiment, said Decision Process Guide may consist of at least one
stored aggregate model and client decision support system, and may be provided
as executable
machine readable code, wherein said code upon execution on a processor causes
said processor to
read only the at least one stored aggregate model provided within said
executable code's data
structure.
In another aspect of this invention, said means to estimate costs of
implementation
project may be determined from combined factor and alternative attributes
within the aggregate
model. In another aspect, said likelihood of success of a project may be
determined from at least
one measure of the degree of success accredited to at least one vendor known
from prior experience
of clients that have used and may still be using said vendor. Said at least
one measure of the degree
of success may be estimated in terms of numeric values representing the
difference between
targeted deliverables - representing a target alternative - in terms of
functionality and solution
quality, and estimates of actual functionality and quality of said
deliverables. Said difference may
be interpreted as a vendor success risk.
Said example illustrates an advisory service process in a procurement
procedure for
an ERP system. Said process can be applied to any procurement process, and can
further include
any decision process open to a systematic decision making process. Said
systematic decision
making processes can include, and may not necessarily limited to, hiring,
performance
CA 02258383 1999-O1-08
_77-
measurement, systems analysis, project performance measurement, medical
decision making, and
strategic planning.
In another embodiment of the process, said tailored model may be stored on at
least
one storage medium such as a Winchester hard disk drive 64 or CD-ROM 68, and
may be read by at
least one special purpose processor 2004 in figure 27, said processor thereby
caused to read said
model by machine readable code executing on said special purpose processor
2004. Said processor
may receive input to from sensors 2001, 2002. For example, in a nuclear power
plant such sensors
may read temperature of coolant. In another embodiment, said sensors may
measure coolant fluid
flow. Said sensors may thereby be any sensor generating a calibrated signal as
may be determined
by one knowledgeable in the art, and said signal converted to a rating value
of significance to the
decision model, where said signal may be represented by a criterion within the
decision model.
From data that may be provided by former tests and similar active sites, said
weights in regard to a
critical situation developing may be determined. For example, sensor 1 (2001 )
near the reactor may
provide temperature readings where it is known that such readings indicate a
likely reactor failure.
Sensor 2 (2002) is further away, and its readings toward a reactor failure may
not be so significant.
Sensor 1 is therefore given for a condition (alternative) called FAILURE
CONDITION A a weight
significantly higher than that of sensor 2 - i.e the local weight of sensor 1
may be much larger that
of sensor 2. If the importance of sensor 1 is determined as ten times the
importance of sensor 2 in
regard to FAILURE CONDITION A (specified as an alternative with respect to the
Goal of
"Reactor Failure"), then the local weight of sensor 1 is 2012 and is
determined to be (w j
=1/11=9.09%), and that of sensor 2 is (wz =10/11=90.91%) 2013. Thence the
sensors, grouped
under the criterion "Temperature Sensors" in the model 2009, may be provided
ratings according to
the temperature signal. Said ratings and weights may then be aggregated into
the hidden node
"Temperature Sensors", 2011 which represents the aggregated signal utility
value from special
processor 2004. In one embodiment of said process, the aggregated node
attributes may be
determined by a machine readable code executed on special purpose processor
2004, said
aggregated attributes may then be transmitted to the higher level general
processor 2007, where said
aggregated attribute may be displayed to an operator on a graphic display
device 2008, where only
node 2011 may be seen. Said device and attribute data may be aggregated
information indicating for
example the status of the processor determined worst temperature reading by
sensors under the
CA 02258383 1999-O1-08
_7g_
hidden node "Temperature Sensors" 2011. In the event the aggregated value
exceeds a preset
threshold (said threshold may be stored locally in local storage device 2003
and read by the special
purpose processor 2004), a signal may be generated by said general purpose
processor, and said
signal causes machine readable code to execute on a special purpose or general
purpose computer
2007, causing said hidden node to expand and expose the underlying data of
node 2011, said
underlying data may then be displayed to at least one operator, and may cause
the generation of at
least one signal to indicate that at least one threshold may have been
exceeded.
In another embodiment, said weights 2012 and 2013 may be changed to cause a
new
perspective to be examined. For example, said temperature sensors in respect
to calibration signals
may be measured against a target set of values, and the importance of said
calibration may change.
For example, sensor 1 may be calibrated against sensor 2, since sensor 2 is
amenable to human
access, and hence the importance of calibration of sensor 2 may be four times
that of sensor 1,
thence the local weights of the two may be sensor 1 80%, and sensor 2 20%, in
confidence of the
calibration performance of the system. Hence, the action of selecting
CALIBRATION
PERFORMANCE STATUS as the alternative, causes said processor to select and
tailor said model
of system for calibration performance status of the system for alternative
CALIBRATION
PERFORMANCE STATUS, including thereby "Temperature Sensors" criteria, but may
not include
other criteria such as mandatory items with discrete values such as valve
closed/open signals. In
another embodiment of the process, said selection of alternative CALIBRATION
PERFORMANCE STATUS may cause machine readable code to execute on a general
purpose or
special purpose processor, causing said processor to disaggregate hidden nodes
2010 and may cause
additional signals to cause automatic measurement of calibration of said
sensors.
In another aspect of the invention, said signals may cause the enabling of at
least one
decision procedures, said at least one decision procedure being present
according to the
aggregated/disaggregated state of the aggregated model. Said procedures may
for example provide
instructions concerning the current state, and may indicate consequences of
action by means of
dynamic graphs and generation of what-if scenarios.
CA 02258383 1999-O1-08
-79-
In another embodiment of the process, processor 2007 may not have within its
model
the nodes known to 2004. Thence at least one output signal 2006 carries
information pertaining to
said unknown nodes to 2007, causing said processor 2007 to create said nodes
2010 with attributes
that may be contained in said at least one signal 2006. Said newly created
nodes may then be
displayed on the graphical display 2008, and may include at least one
attribute of said created
nodes. In the example, said attributes may include weights related to a
decision or analysis
condition, and may include the temperature readouts of said sensors 2001 and
2002. In another
embodiment, nodes 2001 and 2002 may be processors providing 2004 with
aggregate information
of sensors which are descendant nodes to said nodes 2001 and 2002, and so
forth.
Said data for weights that may represent FAILURE CONDITION A may be
generated from prior data determined from similar systems. In another
embodiment, said data may
be generated by means of simulation procedures run on a general purpose or
special purpose
processor. Additionally, said prior data may be processed by machine readable
code to determine at
least one utility function and at least one rating method for ratings, and
weight of importance
toward, for example, the goal of CALIBRATION PERFORMANCE STATUS of the system.
