Aucun titre de diapositive

1
OntoKADS A core ontology to develop expertise
models of the CommonKADS methodology

• Bruaux S. Kassel G.
• LaRIA University Jules Verne of Picardie -
  FRANCE
• Workshop on Core Ontologies in Ontology
  Engineering
• Held in conjunction with EKAW 2004
• 8th October 2004

Laboratory for Research in Computer Science
Amiens (FRANCE) http//www.laria.u-p
icardie.fr
2
Broad outlines
?1. Introduction ?2. The OntoKADS methodology
overview ?3. The OntoKADS ontology
overview ?4. The kernel of the OntoKADS
ontology ?5. Conclusion and perspectives
?1. Introduction ?2. The OntoKADS methodology
overview ?3. The OntoKADS ontology overview ?4.
The kernel of the OntoKADS ontology ?5.
Conclusion and perspectives
3
Introduction (1/7)
Motivations

• Previous works
• Modelling 3 tasks of simulation codes
  calibration
• Bruaux, 2002
• Modelling a generic task of calibration with
  the
• CommonKADS methodology

• Previous works
• Modelling 3 tasks of simulation codes
  calibration
• Bruaux, 2002

Previous works
4
Introduction (2/7)
The CommonKADS knowledge role primitive
Schreiber al., 1999

• roles are inputs/outputs of inferences
• roles are expressed by means of abstract names 

• roles are inputs/outputs of inferences
• roles are expressed by means of abstract names 

5
Introduction (3/7)
Problem the identification of roles (1/3)
6
Introduction (4/7)
Problem the identification of roles (2/3)
7
Introduction (5/7)
Problem the identification of roles (3/3)

• In short
• Abstract roles
• Concrete roles

Knowledge role
domain concept
8
Introduction (6/7)
First attempts to clarify the meaning of the
notion of role
Reynaud C. and al. (1997). The Notion of Role in
Conceptual Modeling. In Proceedings of the 10th
European Knowledge Acquisition Workshop EKAW97,
San Feliu de Guixolls, Bonn Springer Verlag, p.
221-236. Kassel G. (1999). PHYSICIAN is a role
played by an object, whereas SIGN is a role
played by a concept. In Proceedings of the
IJCAI99 Workshop on Ontologies and
Problem-Solving Methods Lessons Learned and
Future Trends, Stockholm (Sweden), August 2,
1999, p. 6-1-6-9.
9
Introduction (7/7)
Using a formal framework to precise the meaning
of CommonKADS modelling primitives

• The ontology of Particulars DOLCE Masolo al.,
  2003, which contains top-level concepts and
  relations
• e.g. PHYSICAL OBJECT, EVENT, ParticipatesIn,
• IsAgentOf
• The Formal Ontology of Properties Guarino
  Welty, 2000, which introduces meta-concepts
  
• e.g. SORTAL, FORMAL ROLE, TYPE, QUASI-TYPE

• The ontology of Particulars DOLCE Masolo al.,
  2003, which contains top-level concepts and
  relations
• e.g. PHYSICAL OBJECT, EVENT, ParticipatesIn,
• IsAgentOf

10
Broad outlines
?1. Introduction ?2. The OntoKADS methodology
overview ?3. The OntoKADS ontology
overview ?4. The kernel of the OntoKADS ontology
?5. Conclusion and perspectives
11
The OntoKADS methodology

Overview
First step using OntoKADS to build an
application ontology
Problem-solving-driven application ontology
12
The OntoKADS methodology

Overview
Second step translating the application ontology
into a CommonKADS expertise model
13
Broad outlines
?1. Introduction ?2. The OntoKADS methodology
overview ?3. The OntoKADS ontology
overview ?4. The kernel of the OntoKADS
ontology ?5. Conclusion and perspectives
14
The OntoKADS ontology
Overview (1/7)
DOLCE in short (1/2)
15
The OntoKADS ontology
Overview (2/7)
DOLCE in short (2/2)
The relation of participation PC(x,y,t) ?def.
x (ED) participates in y (PD) during t
? e.g. the AUTHOR (APO) of an article
PARTICIPATES in the WRITING (AC) of the
article
16
The OntoKADS ontology
Overview (3/7)
Using the Formal Ontology of Properties in short
(1/2)
17
The OntoKADS ontology
Overview (4/7)
Using the Formal Ontology of Properties in short
(2/2)

• The relation of classification
• Cf(x,y,t) ?def. The type x classifies
  the instance y during t
• e.g. SABINE (instance) is the AUTHOR (concept)
  of the article
• The AUTHOR (concept) is
  ANTI-RIGID (meta-concept)

18
The OntoKADS ontology
Overview (5/7)
The problem-solving sub-ontology an extension of
DOLCE

• The problem-solving ontology implements two main
  categories of entities
• acts of Reasoning intervening in problem-solving
  situations
• - e.g. Diagnosing, Calibrating
• entities intervening in these Reasonings
• - e.g. Person, KnowledgeExpression, Agent, Data

• The problem-solving ontology implements two main
  categories of entities
• acts of Reasoning intervening in problem-solving
  situations
• - e.g. Diagnosing, Calibrating

• The problem-solving ontology implements two main
  categories of entities

19
The OntoKADS ontology
Overview (6/7)
The sub-ontology of meta-concepts

• The ontology of meta-concepts implements the
  CommonKADS modelling primitives to classify
  the OntoKADS concepts

• The ontology of meta-concepts implements the
  CommonKADS modelling primitives to classify
  the OntoKADS concepts

• CF(Task, diagnosis, t), CF( Inference, cover,
  t),
• CF(KnowledgeRole, sign, t), CF(DomainConcept,
  car, t)

20
The OntoKADS ontology
Overview (7/7)
Structure of the OntoKADS ontology
DOLCE __
Particular
OntoKADS __
Perdurant
Endurant
Domain Concepts
Domain Concepts
Roles
Action
Communicating
Reasoning
Tasks, Inferences
Transfert Functions
21
Broad outlines
?1. Introduction ?2. The OntoKADS methodology
overview ?3. The OntoKADS ontology
overview ?4. The kernel of the OntoKADS
ontology ?5. Conclusion and perspectives
22
The kernel of the OntoKADS ontology The
problem-solving ontology
The sub-ontology of expressions of knowledge (1/5)
23
The kernel of the OntoKADS ontology The
problem-solving ontology
The sub-ontology of expressions of knowledge (2/5)
24
The kernel of the OntoKADS ontology The
problem-solving ontology
The sub-ontology of expressions of knowledge (3/5)
25
The kernel of the OntoKADS ontology The
problem-solving ontology
The sub-ontology of expressions of knowledge (4/5)
26
The kernel of the OntoKADS ontology The
problem-solving ontology
The sub-ontology of expressions of knowledge (5/5)
ED
DOLCE __
IDA __
POB
NPOB
Inscription
ContentBearingObject
Content
Discourse
LinguisticObject
Concept
Proposition
Term
Information
IconicObject
Assertion
Text
Hypothesis
Complaint
27
The kernel of the OntoKADS ontology The
problem-solving ontology
The sub-ontology of participative roles
28
The kernel of the OntoKADS ontology The ontology
of meta-concepts
Ontological definition of knowledge roles (1/4)

• Three meta-concepts of the Formal Ontology of
  Properties Guarino Welty, 2000 are
  interesting for our study
• role an anti-rigid concept dependent of an
  external entity
• formal role a role lacking an identity
  criterion
• e.g. PATIENT, INSTRUMENT
• material role a role carrying an identity
  criterion
• e.g. STUDENT

29
The kernel of the OntoKADS ontology The ontology
of meta-concepts
Ontological definition of knowledge roles (2/4)
Definition of a MATERIAL ROLE example of the
EMPLOYEE concept
(TYPE) PERSON
(FORMAL ROLE) HUMAN RESOURCE
(MATERIAL ROLE) EMPLOYEE
30
The kernel of the OntoKADS ontology The ontology
of meta-concepts
Ontological definition of knowledge roles (3/4)

• By analogy with the previous definitions, we
  define three notions of "knowledge role"
• KnowledgeRole a role dependent on a Reasoning
• FormalKnowledgeRole a KnowledgeRole lacking an
  identity criterion (e.g. CalibrationData,
  DiagnosisResult)
• MaterialKnowledgeRole a FormalKnowledgeRole
  carrying an identity criterion, which it inherits
  from a Type constrained to be a Proposition
  (e.g. CodeToCalibrate)

31
The kernel of the OntoKADS ontology The ontology
of meta-concepts
Ontological definition of knowledge roles (4/4)
ED
OntoKADS __
(IDA __)
(Content)
(Proposition)
Data
Result
Hypothesis
Code
CalibrationData
DiagnosisResult
CodeToCalibrate
DiagnosisHypothesis
32
Broad outlines
?1. Introduction ?2. The OntoKADS methodology
overview ?3. The OntoKADS ontology overview ?4.
The kernel of the OntoKADS ontology ?5.
Conclusion and perspectives
33
Conclusion and perspectives
Synthesis of our main contributions (1/2)

• The OntoKADS ontology has led us to revisit
  CommonKADS modelling primitives in particular,
  the KnowledgeRole meta-concept differs from the
  meaning given to the knowledge role notion in
  CommonKADS.
• The participants in Reasonings (Tasks) are not
  objects or state of objects (DomainConcepts) but
  Propositions (KnowledgeRole) having
  DomainConcepts as subjects.

• The OntoKADS ontology has led us to revisit
  CommonKADS modelling primitives in particular,
  the KnowledgeRole meta-concept differs from the
  meaning given to the knowledge role notion in
  CommonKADS.

34
Conclusion and perspectives
Synthesis of our main contributions (2/2)

• Two categories of KnowledgeRoles
• FormalKnowledgeRoles, referring to particular
  Reasonings
• ?e.g. CalibrationData, DiagnosisResult
• MaterialKnowledgeRoles, referring to particular
  Reasonings
• and to particular Contents
• ?e.g. CodeToCalibrate, CalibratedCode

Two categories of KnowledgeRoles

• Two categories of KnowledgeRoles
• FormalKnowledgeRoles, referring to particular
  Reasonings
• ?e.g. CalibrationData, DiagnosisResult

35
Conclusion and perspectives
Future works

• Works presented here are progressing in three
  main directions
• the evaluation of the CommonKADS primitives to
  specify problem-solving methods in particular,
  force the methods Inputs/Outputs to be
  KnowledgeRoles would have some consequences on
  these methods
• the integration in OntoKADS of all generic
  Reasonings identified in analytic and synthetic
  Reasonings
• the devolopment of a software, at a conceptual
  level, based on the TERMINAE platform to support
  the OntoKADS methodology

Works presented here are progressing in three
main directions

• Works presented here are progressing in three
  main directions
• the evaluation of the CommonKADS primitives to
  specify problem-solving methods in particular,
  force the methods Inputs/Outputs to be
  KnowledgeRoles would have some consequences on
  these methods

• Works presented here are progressing in three
  main directions
• the evaluation of the CommonKADS primitives to
  specify problem-solving methods in particular,
  force the methods Inputs/Outputs to be
  KnowledgeRoles would have some consequences on
  these methods
• the integration in OntoKADS of all generic
  Reasonings identified in analytic and synthetic
  Reasonings

36
(No Transcript)