MMF Ontology Registration

1
MMF Ontology Registration

• He Keqing and OKABE, Masao
• Project editor
• MMF Ontology Registration
• ISO/IEC JTC1 SC32/WG2
• Ad hoc Liaison Meeting, XMDR, MMF and ODM
• 2005/01/28

2
About this document

• All the materials in this document are based on
  the discussion among and prepared by all the
  active members of ISO/IEC JTC1 SC32/WG2 MMF
  Ontology Registration Project.
• Japan
• Hajime Horiuchi (Tokyo International Univ.)
• Masao Okabe (Project editor, TEPCO)
• Masaharu Obayashi (K-three)
• China
• He Keqing (Project editor, SKLSE, Wuhan Univ.)
• He Yangfan (SKLSE, Wuhan Univ.)
• Wang Chong (SKLSE, Wuhan Univ.)
• Korea
• Doo-Kwon Baik (Korea Univ.)
• Sam Oh (Sungkyunkwan Univ.)

3
Contents

• Basic principle of MMF family
• Outline of MMF Ontology Registration structure
• Relation to ODM
• Evolution management for interoperability
• Relation to MMF Core

4
Basic principle of MMF family

• MMF family do not intend to store a target
  (model, ontology etc.) itself.
• MMF family stores only the administrative
  information of the target and a reference to it
  with a proper granularity

5
Why so, in terms of MMF Ontology Registration

• There are and will be various ontolgies in
  various standardized ontology description
  languages.
• For example, Galen(in OWL), SNOMED(in DL), SUO(in
  KIF(CL)), Italian Opera(in Topic Map)
• If MMF Ontology Registration contains the
  ontology itself, it cannot be independent of the
  structure of its language and difficult to be
  applied to various ontologies in various
  languages.
• On the other hand, the objectives of MMF Ontology
  Registration is to promote interoperability among
  ontology-based application systems in no matter
  what languages their ontologies are described.
• So, it should reflect only a common basic
  structure of ontologies and store their
  administrative information only so that it can be
  applied to various ontologies in various
  languages.

6
Common basic structure of ontology

• A very simplified but common three granularity
  level structure is

• An ontology consists of ontology components
• (or sentences, pieces of knowledge)
• e.g. Example_Ontology consists of
• ?Buyer? ?has.Creditrating(Tony)
• Buyer(Tony)
• Creditrating(Credit-A)
• A ontology component consists of
• atomic constructs (symbol, or concepts and
  objects etc. or predicates and individuals etc.).
• e.g. ?Buyer? ?has.Creditrating(Tony) consists of
• Buyer
• has
• ? , ? , ?

• Creditrating
• Tony

7
Common structure of ontology description languages

• Almost any FOLs have these hierarchies.

expression
This is Atomic Ontology Construct
This is Ontology Component
symbol
term
sentence (in a broad sense)
Atomic term
composite term
sentence (or clause) (in a narrow sense)
definition
logical symbol (in a broad sense)
This is Atomic Ontology Construct in MMF
Ontology Registration
non logical symbol
variable
logical symbol (in a narrow sense)
predicate
individual (or object)
unary predicate (or concept)
N-nary predicate (or role, relation)
sentence letter (o-ary predicate)
8
Outline of MMF Ontology Registration structure
Administrative information corresponding to
Ontology

• e.g.
• Administrative information of Example_Ontology
• e.g.
• Administrative information of each
• ?Buyer? ?has.Creditrating(Tony)
• Buyer(Tony)
• Creditrating(Credit-A)
• e.g.
• Administrative information of each
• Buyer
• has
• NoteLogical symbols such as ? , ? , ? are
  ignored.

Administrative information corresponding to
Ontology Component
Administrative information corresponding to
Atomic Ontology Construct

• Creditrating
• Tony

9
Relation between MMF Ontology Registration and
ODM(1)
MMF Ontology Registration
Ontology

• Hereinafter called
• Ontology etc.
• just for simplification.

Ontology Component
Atomic_Onto_Construct
We believe that ODM with XMI will provide a
suitable interface to MMF Ontology Registration.
ER Metamodel
UML2 Metamodel
SCL Metamodel
OWL/RDFS Metamodel
TM Metamodel
DL Metamodel
ODM
We believe that an ontology can be stored in a
ODM metamodel without any loss of information.
Ontology described in OWL/RDFS
10
Example 1 from ODM revised submission B.7.1
MMF Ontology Registry
Example1Ontology administrative information
e1Onto_Component administrative information
e5Onto_Component administrative information
PersonalCarAtomic_Onto_Construct administrative
information
CarAtomic_Onto_Construct administrative
information
PersonAtomic_Onto_Construct administrative
information
ownsAtomic_Onto_Construct administrative
information
VehicleAtomic_Onto_Construct administrative
information
11
Relation between MMF Ontology Registration and
ODM(2)
MMF Ontology Registration
ltnowgt
ODM (revised submission)
SCL Metamodel
UML2 Metamodel
ER Metamodel
OWL/RDFS Metamodel
TM Metamodel
DL Metamodel
Ontology described in OWL/RDFS
MMF Ontology Registration
ltbeforegt
ODM Mapping
ER Metamodel
UML2 Metamodel
SCL Metamodel
ODM (preliminary revised submission)
OWL/RDFS Metamodel
TM Metamodel
DL Metamodel
Ontology described in OWL/RDFS
12
MMF Ontology Registration from the point of
evolution management for interoperability

• A local ontology selects
• its knowledges from
• a valid variant source ontology.

• In a valid variant source ontology,
• all knowledges in a source ontology is valid and
  there may be some new knowledges
• and some name changes for a local ontology.

• Local ontology
• localized ontology for some application system
  based on source ontologies
• relatively unstable and evolves autonomously

• Source Ontology
• standardized ontology
• for some business domain
• relatively stable

13
Example 2 example description

• Source ontologies
• SO1
• ?Buyer? ?has.Creditrating
• Buyer(Anthony)
• Creditrating(Credit-A)
• has(Anthony, Credit-A)
• Note This is described in DL
• Local Ontology
• LO1
• Buyer(Tony)
• Creditrating(Credit-A)
• has(Tony, Credit-A)
• hasProblem(Tony, A)
• About(A, Credit-A)
• Note
• The name Anthony in SO1 and SO2 is changed to
  Tony in LO1
• A new knowledge About(A, Credit-A) is added.

• SO2
• hasProblem(Anthony, A)
• Email(B)
• Send(Anthony, B, Jerry)
• Note
• This can be understood as in KIF etc. but cannot
  be in DL nor OWL because Send is 3-aray
  predicate.
• LO2
• Buyer(Anthony)
• Email(B)
• Send(Anthony, B, Jerry)
• About(B, A)
• Note
• A new knowledge About(B,A) is added locally.

14
Example2Without (Valid Variant of ) Source
Ontology
Agent A of the application system based on LO1
Agent B of the application system based on LO2
Tell me to whom Tony sent an e-mail?
Local Ontology LO1
Local Ontology LO2
Tony??? I do not know Tony.

• Buyer(Tony)
• About(A, Credit-A)
• has(Tony, Credit-A)
• hasProblem(Tony, A)
• Creditrating(Credit-A)

• Buyer(Anthony)
• Email(B)
• Send(Anthony, B, Jerry)
• About(B, A)

• What is the worse,
• it is difficult for agent A to find agent B who
  has the answer.

15
Example2 with (Valid Variant of ) Source Ontology
Agent of MMF Ontology Registration tells agent A
that agent B can answer it.
Agent A of the application system based on LO1
Agent B of the application system based on LO2
Tell me to whom Tony sent an e-mail?
Local Ontology LO1
Local Ontology LO2
Hmm.. Tony is Anthony. So, the answer is to
Jerry.

• Buyer(Tony)
• About(A, Credit-A)
• has(Tony, Credit-A)
• hasProblem(Tony, A)
• Creditrating(Credit-A)

• Buyer(Anthony)
• Email(B)
• Send(Anthony, B, Jerry)
• About(B, A)

16
Example2Object Diagram of MMF Ontology
Registration
ltSource_Ontologygt
ltVVSource_Ontologygt
ltAtomic_Onto_Constructgt
SO1
VVSO1
Creditrating
ltOnto_Componentgt
ltLocal_Ontologygt
Buyer
?Buyer? ?has.Creditrating
LO1
Anthony
Buyer(Tony)
Buyer(Anthony)
Tony
Creditrating(Credit-A)
Credit-A
has(Tony, Credit-A)
has(Anthony, Credit-A)
has
About(A, Credit-A)
About
hasProblem
About(B, A)
A
hasProblem(Tony, A)
hasProblem(Anthony, A)
Email
Email(B)
B
LO2
Send(Tony, B, Jerry)
Send(Anthony, B, Jerry)
Send
Jerry
VVSO2
SO2
17
Example 3SUMO

• Source_Ontology
• An administrative information corresponding to
  SUMO at http//virtual.cvut.cz/kifb/en/
• Ontology_Component
• An administrative information corresponding to
• (gt (and (instance ?LANG AnimalLanguage) (agent
  ?PROC ?AGENT) (instrument ?PROC ?LANG)) (and
  (instance ?AGENT Animal) (not (instance ?AGENT
  Human)))), etc.
• This is in KIF(CL) and in English, If lang is an
  instance of animal language and proc is an agent
  of agent and lang is an instrument for proc, then
  agent is an instance of animal and agent is not
  an instance of human.
• Atomic_Onto_Construct
• An administrative information corresponding to
• instance, agent, instrument, instance, Note
  these are binary relations.
• AnimalLanguage, Animal, Human, Note these are
  concepts.
• Note ?LANG, ?PROC, ?AGENT are just variables
  and not individuals.

18
Example 4 OWL Wine Ontology (1/2)

• Source_Ontolgy
• An administrative information corresponding to a
  whole ontology wine.xml at http//www.w3.org/TR
  /2003/PR-owl-guide-20031209/wine
• Ontology_Component
• An administrative information corresponding to
• ltowlClass rdfID"WhiteWine"gt
• ltowlintersectionOf rdfparseType"Collection"gt
  
• ltowlClass rdfabout"Wine" /gt
• ltowlRestrictiongt
• ltowlonProperty rdfresource"hasColor" /gt
  
• ltowlhasValue rdfresource"White" /gt
• lt/owlRestrictiongt
• lt/owlintersectionOfgt
• lt/owlClassgt ,
• ltowlObjectProperty rdfID"hasVintageYear"gt
• ltrdftype rdfresource"owlFunctionalProperty"
  /gt
• ltrdfsdomain rdfresource"Vintage" /gt
• ltrdfsrange rdfresource"VintageYear" /gt
• lt/owlObjectPropertygt , etc

19
Example 4 OWL Wine Ontology (2/2)

• Atomic_Onto_Construct
• An administrative information corresponding to
• WhiteWine,
• Collection,
• Wine,
• hasColor,
• White,
• hasVintageYear,
• FunctionalProperty,
• Vintage,
• VintageYear, etc

20
Relation between MMF Ontology Registration and
Core

• This is a very simplified diagram and may need
  some refinements.

21

• Thank you.
• Your support would be highly appreciated!

22

• ?????????????????????????????????