Making Better Strategic Making Better Strategic Decisions,Decisions,

by Asking If It Is Going To Get by Asking If It Is Going To Get Better or WorseBetter or Worse

- the SOFI Method - the SOFI Method and the System to Implement itand the System to Implement it

13-Dec-2007

TedGordon, PeterYim, AdamCheyer, DeniseBedford, PatCassidy & KenBaclawski

(v 0.20)

NASA-Ontolog-KMWG “OKMDS” mini-series

Ontology in Knowledge Management and Decision Support

(OKMDS)

AbstractAbstract• Simply ask "If It Is Going To Get Better or Worse", and Simply ask "If It Is Going To Get Better or Worse", and

then just go with the decision that would get things then just go with the decision that would get things better for us! This almost seems intuitively obvious ... better for us! This almost seems intuitively obvious ... until we delve seriously into the issue, and start until we delve seriously into the issue, and start wondering ...wondering ...

• see: see: http://ontolog.cim3.net/cgi-bin/wiki.pl?ConferenceCall_2007_12_13#nid15SG

• ... The team is on the R&D to cast the SOFI work into a ... The team is on the R&D to cast the SOFI work into a collaborative open knowledge system that brings to collaborative open knowledge system that brings to bear some of the best legacies passed on to us, as well bear some of the best legacies passed on to us, as well as state-of-the-art approaches in information as state-of-the-art approaches in information management and technology, knowledge management, management and technology, knowledge management, quantitative analysis, qualitative analysis, service quantitative analysis, qualitative analysis, service oriented architecture, web 2.0, Delphi, as well as oriented architecture, web 2.0, Delphi, as well as semantic technologies. The use of ontologies and semantic technologies. The use of ontologies and ontological engineering will be central to their ontological engineering will be central to their approach in the next phase of the work, that will approach in the next phase of the work, that will enable their tapping into virtual communities and enable their tapping into virtual communities and disparate systems to come up with the collective disparate systems to come up with the collective intelligence to provide strategic decision support.intelligence to provide strategic decision support.

OutlineOutline• Introduction Introduction (Peter-3)(Peter-3)

• Futures Research, Normative Futures, Delphi & SOFI Futures Research, Normative Futures, Delphi & SOFI (Ted-(Ted-8)8)

• Core Concepts, Values & the SOFI System Mission Core Concepts, Values & the SOFI System Mission (Peter-4)(Peter-4)

• Phase-1 SOFI System implementation Phase-1 SOFI System implementation (Adam/Peter-4)(Adam/Peter-4)

• Driving the Next Phase with ontology engineering:Driving the Next Phase with ontology engineering: Quantitative Decision Support Quantitative Decision Support (Ken-4)(Ken-4)

toward semantic interoperability through a toward semantic interoperability through a foundational ontology foundational ontology (Pat-4)(Pat-4)

a collaborative knowledge environment a collaborative knowledge environment (Adam-4)(Adam-4)

• Architecting a co-evolving human-machine system for Architecting a co-evolving human-machine system for collective intelligence collective intelligence (Peter/Adam-3)(Peter/Adam-3)

• Making Better Strategic Decisions – applications, Making Better Strategic Decisions – applications, scenarios and the potentials of a SOFI system scenarios and the potentials of a SOFI system (Denise/Ted-3/3)(Denise/Ted-3/3)

• Q & A – discussion Q & A – discussion (All)(All)

The Millennium Project The Millennium Project - MP / WFUNA- MP / WFUNA

• The The Millennium ProjectMillennium Project of WFUNA is ... of WFUNA is ... a global participatory futures research think a global participatory futures research think

tank of futurists, scholars, business planners, tank of futurists, scholars, business planners, and policy makers who work for international and policy makers who work for international organizations, governments, corporations, organizations, governments, corporations, NGOs, and universities , started in 1996?NGOs, and universities , started in 1996?

operating as a network of 32? globally operating as a network of 32? globally located nodes located nodes

with contribution in both content and with contribution in both content and methodologymethodology

under the auspices of the World Federation under the auspices of the World Federation of United Nations Associations (WFUNA)of United Nations Associations (WFUNA)

• not to be confused with the UN/Millennium Projectnot to be confused with the UN/Millennium Project

State of the Future Index State of the Future Index - SOFI- SOFI

• ....

Full Implementation of aFull Implementation of a SOFI SOFI SystemSystem

• ....

Futures Research, Futures Research, Normative Futures, Normative Futures,

Delphi Delphi & SOFI& SOFI

Ted Gordon, The Millennium Project / WFUNA

Purposes of Futures Purposes of Futures ResearchResearch

To help understand what could be, To help understand what could be, what might be, and what ought to what might be, and what ought to be.be.

To discover and understand threats To discover and understand threats and opportunities.and opportunities.

To develop more creative strategies To develop more creative strategies and evaluate proposed actions.and evaluate proposed actions.

To create and share normative To create and share normative visions visions

The value of futures research lies The value of futures research lies less in its forecasting accuracy, than less in its forecasting accuracy, than in its use in identifying and in its use in identifying and assessing new possibilitiesassessing new possibilities

Philosophical AssumptionsPhilosophical Assumptions

The future can be shaped by policyThe future can be shaped by policy There is a range of possible futuresThere is a range of possible futures Policy consequences can be Policy consequences can be

systematically exploredsystematically explored Exact foreknowledge is not possible, Exact foreknowledge is not possible,

but probabilities can be assessedbut probabilities can be assessed There is a component of the future There is a component of the future

that is unknowablethat is unknowable The mind uses “what if” outcome The mind uses “what if” outcome

scenariosscenarios

Key Current MethodsKey Current Methods

TaxonomyTaxonomyNormative vs. Normative vs. exploratoryexploratoryQuantitative Quantitative vs. qualitativevs. qualitative

Common Common techniquestechniques

Time SeriesTime Series Delphi Delphi

ScenariosScenariosModeling Modeling MonitoringMonitoring

FrontiersFrontiers

MethodsMethodsTrend Impact AnalysisTrend Impact AnalysisSOFISOFIRTDelphiRTDelphiScenarios using expert groupsScenarios using expert groupsRobust decision makingRobust decision making

ConceptsConceptsReducing the unknownReducing the unknownA new decision scienceA new decision science

Time Series: Trend Impact Time Series: Trend Impact AnalysisAnalysis

Infant Infant Mortality Mortality RateRate

Very long term, low cost contraceptives.Very long term, low cost contraceptives.

China and India becomes 50% urbanizedChina and India becomes 50% urbanized

Social marketing improves health practicesSocial marketing improves health practices

Standard vaccinations of all childrenStandard vaccinations of all children

Improved maternal healthcare, nutritionImproved maternal healthcare, nutrition

Literacy rate of women reaches 85%Literacy rate of women reaches 85%

Number of poor people growsNumber of poor people grows

Drop in number of people w/o safe waterDrop in number of people w/o safe water

1983 1988 1993 1998 2003 2008 2013

Year

20

30

40

50

60

70

80

90

100

Extrapolation

Low

Median

High

Infant Mortality; Deaths per 1,000 Live Births

Time Series: Trend Impact Time Series: Trend Impact AnalysisAnalysis

DelphiDelphi

Conference room problemsConference room problems

Now the “classic” means for Now the “classic” means for combining expert opinionscombining expert opinions

AnonymityAnonymity

FeedbackFeedback

The RAND 1964 S&T StudyThe RAND 1964 S&T Study

1

Infant mortality (deaths per 1,000 births)World (1983)= 89.1World (2005)= 50.0Spain (2003)= 4Liberia (2003)= 157Link go

Best plausible value in 2016? YOUR ANSWER :The average group answer: 23.4 Respondents: 7 DIFFERENT THAN THE AVERAGE Reasons click here GO:

Real Time Delphi

Example of SOFI (State of the Future Index) variables:Infant mortalityFood availabilityGNP per capita Access to fresh waterCO2 emissionsLiteracyWarsAIDS deathsTerrorist attacksDebt ratioUnemploymentCalories per capitaHealth careForest landsRich poor gap…

Global SOFI 2006: Global SOFI 2006: 20 Variables20 Variables

Comparison of SOFI's

0.5

0.6

0.7

0.8

0.9

1

1.1

1.2

1980 1985 1990 1995 2000 2005 2010 2015 2020

2001

2002

2003

2004

2005

Core Concepts, Values Core Concepts, Values and and

the SOFI System Missionthe SOFI System Mission

Peter Yim, CIM3

Core Concepts & ValuesCore Concepts & Values driving this workdriving this work

• The SOFI Methodology The SOFI Methodology

• The "Bootstrap" ParadigmThe "Bootstrap" Paradigm

• Building a knowledge environment to serve a Building a knowledge environment to serve a wide variety of applicationswide variety of applications

• Collaboration among participants and Collaboration among participants and stakeholders in the form of virtual stakeholders in the form of virtual communities. communities.

• The notion of "Openness"The notion of "Openness"

• Driving interoperability through ontology Driving interoperability through ontology engineeringengineering

• Integrating through a service oriented Integrating through a service oriented architecturearchitecture

• Tapping into a collective intelligence for Tapping into a collective intelligence for decision supportdecision support

Phase-1 SOFI System Phase-1 SOFI System implementationimplementation

Peter Yim (CIM3), Adam Cheyer (SRI), et al.

Background Background (1)(1)

• Two Concepts guiding this work:Two Concepts guiding this work:o Futures Research leading to normative futuresFutures Research leading to normative futureso AugmentedAugmented Collaboration towards Collective Collaboration towards Collective

IntelligenceIntelligence

• The Internet moves on: The Internet moves on: o from the research and academic networkfrom the research and academic networko to being a publishing mediato being a publishing mediao to Transactions and Web Servicesto Transactions and Web Serviceso on to being a knowledge media - the Semantic on to being a knowledge media - the Semantic

WebWeb

• This is where we begin – where, hopefully, This is where we begin – where, hopefully, human systems and tools systems may co-human systems and tools systems may co-evolveevolve

The System The System (1)(1)

• ArchitectureArchitecture• The Knowledge Base/EngineThe Knowledge Base/Engine• The SOFI ComputerThe SOFI Computer• The Community of Practice The Community of Practice

(CoP) Collaborative Work (CoP) Collaborative Work EnvironmentEnvironment

• The User & the User InteractionThe User & the User Interaction

ArchitectuArchitecturere

The User InterfaceThe User Interface

The Next Phase ...The Next Phase ...driving the SOFI Systemdriving the SOFI System

with semantics with semantics and ontology engineeringand ontology engineering

( ... do an intro here, or insert one additional slide after this to do it)

Quantitative Decision Quantitative Decision SupportSupport

Ken Baclawski, Northeastern University

Decision AnalysisDecision Analysis

• Important part of policy and Important part of policy and development processes.development processes.

• Formal annotation of decisions and Formal annotation of decisions and their analyses can have many benefits.their analyses can have many benefits.– Integration with the processIntegration with the process– Recognition of need to reconsider when Recognition of need to reconsider when

circumstances evolvecircumstances evolve– Decisions can be delayedDecisions can be delayed– Decisions can be reused for other Decisions can be reused for other

situationssituations• An annotated decision is called a An annotated decision is called a

rationalerationale

Rationale OntologyRationale Ontology

Rationale

Artifact

Choice

Decision Analysis

Informal Discussion

Influence Diagram

Decision Tree

Decision Table

analysis

issue

Evidence

evidence*

criterion*

decision

alternative*

isa isa

isa

isa

isa

isa

Process ModelProcess Model

Policy Decision ExamplePolicy Decision Example

Investment Level in Brain Health Intervention Techniques

Ageism

Fertility Rate

Age of Population

Ethical Concerns

Standard of Living

Brain Health Level

Brain Health Policy Rationale

criterion

alternative

analysis

Census data

evidence

Brain Health Issue

issue

affects

Toward semantic Toward semantic interoperability through a interoperability through a

foundational ontologyfoundational ontology

Dr. Patrick Cassidy, Micra

31

What is a “Defining What is a “Defining Vocabulary”?Vocabulary”?

• For lexicographers, a controlled list of For lexicographers, a controlled list of words which are the only words allowed to words which are the only words allowed to be used in creating definitions (e.g. in be used in creating definitions (e.g. in LDOCE).LDOCE).– makes definitions easier to understand, makes definitions easier to understand,

especially for learners of a language (like especially for learners of a language (like dumb computers)dumb computers)

• For Ontologists, the “Conceptual Defining For Ontologists, the “Conceptual Defining Vocabulary” (CDV) is the set of basic Vocabulary” (CDV) is the set of basic concepts (and their ontological concepts (and their ontological representations) which are necessary and representations) which are necessary and sufficient to specify the meanings of all sufficient to specify the meanings of all other concepts (or terms) by combinations other concepts (or terms) by combinations of the basic concepts – of the basic concepts – a special type of a special type of “Foundation Ontology”.“Foundation Ontology”.

How Will a Conceptual Defining How Will a Conceptual Defining vocabulary Enable Semantic vocabulary Enable Semantic

Interoperability?Interoperability?• Any two ontologies developed by Any two ontologies developed by

specifying the meanings of terms or specifying the meanings of terms or ontology elements (types, relations, ontology elements (types, relations, functions, rules) as combinations of the functions, rules) as combinations of the foundation ontology elements can foundation ontology elements can automatically create a common merged automatically create a common merged ontology by sharing specifications of ontology by sharing specifications of ontology elements that are not in the ontology elements that are not in the starting common ontology.starting common ontology.

• The inferences that both systems The inferences that both systems derive from the same data will then derive from the same data will then be identical.be identical.

What Is The Advantage of a What Is The Advantage of a “Conceptual Defining “Conceptual Defining

Vocabulary”?Vocabulary”?• Keep the Foundation Ontology as Keep the Foundation Ontology as

small as possible, to maximize small as possible, to maximize chances of getting agreement among chances of getting agreement among a large user audience.a large user audience.

• Focus on the elements that are Focus on the elements that are essential to support the most basic essential to support the most basic components of human-level components of human-level understanding.understanding.

• Provide all of the elements that are Provide all of the elements that are required to support interoperability, required to support interoperability, and achieve relative stability quickly.and achieve relative stability quickly.

34

How Large is the Conceptual How Large is the Conceptual Defining Vocabulary?Defining Vocabulary?

• Longman’s Dictionary of Contemporary Longman’s Dictionary of Contemporary English (LDOCE) uses about 2000 root English (LDOCE) uses about 2000 root words, some of which are used in more words, some of which are used in more than one sense. With morphological than one sense. With morphological variants (plurals, past tenses), there are variants (plurals, past tenses), there are over 9000 words.over 9000 words.

• For the conceptual defining vocabulary, For the conceptual defining vocabulary, probably at least 4000 senses will be probably at least 4000 senses will be needed.needed.

• The vocabulary will probably grow over The vocabulary will probably grow over time.time.

• But as more fields are defined via the But as more fields are defined via the conceptual defining vocabulary, each new conceptual defining vocabulary, each new field should require a smaller percentage of field should require a smaller percentage of new primitives to be added to the new primitives to be added to the foundation ontology to support the foundation ontology to support the specification of novel concepts.specification of novel concepts.

• The rate at which the need for new The rate at which the need for new primitives will grow will be determined by primitives will grow will be determined by actual experience.actual experience.

How Does the CDV Differ How Does the CDV Differ from other “Upper from other “Upper

Ontologies”?Ontologies”?(1)(1)Primary focus is on the most basic Primary focus is on the most basic

concepts, to maximize agreement and concepts, to maximize agreement and usabilityusability

(2)(2)This focus is achieved by concentrating on This focus is achieved by concentrating on representing the concepts labeled by the representing the concepts labeled by the words of the linguistic defining words of the linguistic defining vocabulary, with tight integration of the vocabulary, with tight integration of the ontology and linguistic vocabulary.ontology and linguistic vocabulary.

(3)(3)The first demonstration application would The first demonstration application would be to enable a basic child’s-level ability to be to enable a basic child’s-level ability to use the English language. Until that is use the English language. Until that is achieved, more specialized applications achieved, more specialized applications are unlikely to approach human-level are unlikely to approach human-level understanding of concepts.understanding of concepts.

A Collaborative A Collaborative Knowledge Knowledge

EnvironmentEnvironment

Adam Cheyer, SRI

(... insert ~4 slides from Adam)

Architecting a co-evolving Architecting a co-evolving human-machine system human-machine system

for collective intelligencefor collective intelligence

Adam Cheyer (SRI), Duane Nickull (Adobe), Peter Yim (CIM3), et al.

( insert ~3 slides from Adam/Duane/Peter ...

o an ontology framework slide o a SOA RM slide o a next phase SOFI System slide )

Making Better Strategic Making Better Strategic Decisions – applications, Decisions – applications,

scenarios and the scenarios and the potentials of a SOFI potentials of a SOFI

systemsystemTed Gordon, MP/WFUNA

&

Denise Bedford, World Bank

Example of SOFI (State of the Future Index) variables:Infant mortalityFood availabilityGNP per capita Access to fresh waterCO2 emissionsLiteracyWarsAIDS deathsTerrorist attacksDebt ratioUnemploymentCalories per capitaHealth careForest landsRich poor gap…

Global SOFI 2006: Global SOFI 2006: 20 Variables20 Variables

Comparison of SOFI's

0.5

0.6

0.7

0.8

0.9

1

1.1

1.2

1980 1985 1990 1995 2000 2005 2010 2015 2020

2001

2002

2003

2004

2005

What Goes Into a What Goes Into a Decision?Decision?

Three questions:Three questions:– What is possible?What is possible?– What is likely?What is likely?– What is desirable?What is desirable?

Futures research helps withFutures research helps with– Identifying goalsIdentifying goals– Assessing prospective policies Assessing prospective policies – Quantifying risksQuantifying risks

Futures research does not help much withFutures research does not help much with– Reducing the unknowableReducing the unknowable– Psychological factorsPsychological factors– Moral and ethical factorsMoral and ethical factors

A New Decision ScienceA New Decision Science

Futures research; foresight; FTAFutures research; foresight; FTA

Intuition, imagination: experience, subtle cluesIntuition, imagination: experience, subtle clues

Psychology; personal utility functionsPsychology; personal utility functions

Balance of risks and rewardsBalance of risks and rewards

Experiments and analogyExperiments and analogy

Understanding innate illogic of human thinkingUnderstanding innate illogic of human thinking

Moral courageMoral courage

(... insert ~3 slides from Denise)

ReferencesReferences• ... add references here... add references here

Q & A Q & A and and

DiscussionDiscussion