various ontologies and mmf ontology registration

40
OFMR200 6 Various ontologies and MMF Ontology Registrati on OKABE, Masao Co-editor, MMF Ontology Registration Project, ISO/IEC JTC1 SC32/WG2 Corporate Systems Department, Tokyo Electric Power Co., Inc. 2006.3.21

Upload: fathia

Post on 30-Jan-2016

39 views

Category:

Documents


0 download

DESCRIPTION

Various ontologies and MMF Ontology Registration. OKABE, Masao Co-editor, MMF Ontology Registration Project, ISO/IEC JTC1 SC32/WG2 Corporate Systems Department, Tokyo Electric Power Co., Inc. 2006.3.21. Introduction. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Various ontologies and  MMF Ontology Registration

OFMR2006

Various ontologies and MMF Ontology Registration

OKABE, MasaoCo-editor, MMF Ontology Registration Project, ISO/IEC JTC1 SC32/WG2

Corporate Systems Department, Tokyo Electric Power Co., Inc.

2006.3.21

Page 2: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

2(無断複製・転載禁止)

IntroductionNowadays, “Ontology” is almost a buzzword and the meaning is very br

oad and is not so clear.

In this presentation, I will present;my personal understanding of ontologies

andhow MMF Ontology Registration* is related to my understanding of o

ntologies.

This topic is very challenging to me. Any comments are very welcome.

* Note: MMF Ontology Registration is a part of mulch part standards ISO/IEC 19763#

# Project leader: Hajime Horiuchi (Tokyo International University, Japan) Its current status is FCD. Co-editors of MMF Ontology Registration are

HE Keqing (SKLSE, Wuhan university, China) OKABE, Masao (Tokyo Electric Power Co., Inc., Japan)

Page 3: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

3(無断複製・転載禁止)

Outline

1. My personal understanding of ontologies

2. Various kinds of ontologies

3. Trustiness of ontologies

4. MMF Ontology Registration

Appendix: What TEPCO intends to do

Page 4: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

4(無断複製・転載禁止)

1. My personal understanding of ontologies

2. Various kinds of ontologies

3. Trustiness of ontologies

4. MMF Ontology Registration

Appendix: What TEPCO intends to do

Page 5: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

5(無断複製・転載禁止)

Famous definition of ontology

A famous definition is;

“An ontology is a specification of a conceptualization.”

by T. R. Gruber

at http://www-ksl.stanford.edu/kst/what-is-an-ontology.html

This definition is very abstract, but has a very important point.That is, “a specification of a conceptualization”

Page 6: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

6(無断複製・転載禁止)

A specification of a conceptualizationUsing the famous “Meaning Triangle” by Ogden and Richards,

a specification of a conceptualization can be explained as a image of mapping from UoD to the world of symbols based on thought.

That is, ontology is not a thing in our mind and is an explicit expression in some language (or word).

Referent (in UoD)Symbol

A chef bakes a pancake!

x Chefy Pancake Bake ( x, y )

specification of conceptualization

Thought

Page 7: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

7(無断複製・転載禁止)

Simple use case A simple use case from

7.2 Wine Agent, 7.Usage Example, OWL Web Ontology Language Guide, http://www.w3.org/TR/owl-guide/

Wine Ontology

Wine agent

Tell me a appropriate wine for the dinner.I will serve a special tomato based pasta source with fresh pasta as the main course.

The recommended one is Marietta Zinfandel.

Where can I buy it?

Winerlibrary.com has a sale on it.

Wine agent needs to be able to understand Wine ontology and find a suitable answer.

Page 8: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

8(無断複製・転載禁止)

What is an ontology?

From this simple use case, somewhat more specific definition can be gained, that is,

An ontology is a set of descriptions of the UoD in a formal language so that a computer can understand it and share it.

Then, the question is what it means that “A computer can understand an ontology and share an ontology.”

I think there are two meaningsOne is “Uniqueness of interpretation”.

almost impossibleThe other is “Uniqueness of inference”.

needs to be embodied.

Page 9: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

9(無断複製・転載禁止)

Uniqueness of interpretation

“Uniqueness of interpretation” means that, for example, symbol “Wine” in the Wine ontology cannot be interpreted as other than wine in the UoD.

But, Semantics, Logical Entailment in KIF manual( http://logic.stanford.edu/kif/Hypertext/node13.html )

The goal of knowledge encoding is to write enough sentences so that unwanted interpretations are eliminated. Unfortunately, this is not al

ways possible.

I do not know why it is not always possible, but practically it seems so because, first of all, to define UoD rigorously is almost impossible.

Actually, there may be an interpretation that interprets “Wine” as Japanese sake and “Winegrape” as sake-rice.

Page 10: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

10(無断複製・転載禁止)

Uniqueness of inference (1 of 3)

A little bit different picture of Simple use caseWine

Ontology

MyPortal agent

Tell me a appropriate wine for the dinner.I will serve a special tomato based pasta source with fresh pasta as the main course.

The recommended one is Marietta Zinfandel.

Where can I buy it?

Winerlibrary.com has a sale on it.

Wineagent

How much is it?

Uhm… There are so many questions.Wine ontology is open to all. You can find the answer by yourself.

Page 11: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

11(無断複製・転載禁止)

Uniqueness of inference (2 of 3)

Then,

Wine Ontology

MyPortal agent

Again,Tell me a appropriate wine for the dinner.I will serve a special tomato based pasta source with fresh pasta as the main course.

The recommended one is St. Clement Merlot.

What!!! Isn’t it Marietta Zinfandel???

Uniqueness of inference means that this kind of things never happen.

Page 12: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

12(無断複製・転載禁止)

Uniqueness of inference (3 of 3)

Suppose computer A and B have the same ontology, i.e. the same set of

descriptions of the universe of discourse.

Then, From the ontology, computer A can draw all the inferences that

computer B can draw from the ontology. And vice versa.The inferences that computer A and B draw from the ontology does

not contradict the inferences a human draws from the universe of discourse that the ontology describes.

Page 13: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

13(無断複製・転載禁止)

Requirements of a language for ontologies

Based on the definitions of ontologies above,

there are two requirements of a language for ontologies

First, “Syntax”

so that a computer can analyze the descriptions.

Second, “Formal semantics (or model-theoretic semantics or interpretation conditions etc.)”

that specifies conditions of interpretations and procedures of inference at least sound, usually complete and preferably decidable

Page 14: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

14(無断複製・転載禁止)

Ontology vs. Knowledge base

So, in my formal position, there is no specific distinction between an ontology and a knowledge base.

But, intuitively, there is difference between an ontology and a knowledge base.

An ontology does not lay an emphasis on inference so much as an knowledge base does, but mainly focuses on describing the universe of discourse so far as necessary.

Page 15: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

15(無断複製・転載禁止)

About formal semantics

In my informal position, I accept ontologies that are described in a language that does not have its formal semantics explicitly. So, for example, a model in UML can be an ontology.

It depends on what kind of query (in a broad sense) an ontology accepts whether an ontology needs its formal semantics or not.

If an ontology accepts only usual procedural queries, something like in SQL, then it does not need its formal semantics explicitly.

If an ontology accepts more declarative queries, something like ‘Tell me the suitable wine for the pasta with tomato’, then it needs its formal semantics.

Page 16: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

16(無断複製・転載禁止)

1. My personal understanding of ontologies

2. Various kinds of ontologies

3. Trustiness of ontologies

4. MMF Ontology Registration

Appendix: What TEPCO intends to do

Page 17: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

17(無断複製・転載禁止)

Various kinds of ontologies

Based on my understanding of ontologies above,

there are several axes to classify ontologies,

which are not necessarily uncorrelated,

in addition to “Upper ontology vs. Domain ontology”.

Heavy weight ontology vs. Light weight ontology Ontology of the real world vs. of information systemsOntology on the static aspect vs. on the dynamic aspectOntology with data-flavor vs. with program-flavorOntology for a computer vs. for a human

Page 18: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

18(無断複製・転載禁止)

Heavy weight vs. Light weightHeavy weight ontology

A rich and formal ontology that an inference engine can derive necessary results.

Example “Wine ontology” (and “Food ontology”) from which an agent can derive a suitable w

ine for a dish.–Actual “Wine ontology” at http://www.w3.org/TR/2003/PR-owl-guide-20031209/wine and “F

ood ontology” at http://www.w3.org/TR/2003/PR-owl-guide-20031209/food are not so rich, even though “Food ontology” has several rules on the appropriate combination of food and wine.

Ontologies based on PSL-Core and -Outercore in TC184 or FLOWS-Core in W3CLight weight ontology

A relatively simple ontology that mainly focuses on the relations among concepts.

No explicit distinction from a taxonomy and a thesaurus. Example:

Ontologies in RDF or Topic Maps Traditional terminological medical ontologies such as SNOMED-III Ontologies in MIT Process Handbook

Page 19: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

19(無断複製・転載禁止)

Real world vs. Information systems (1 of 2)The real world

Ontology of the real world

Note: This issue is almost the same as whether class “employee” means an employee in the real world or the objects in the HR systems at an object-oriented analysis.

information systems (specifications)

Ontology of information systems

modeling for

information system

stan

ds

for

specifications of conceptualizations

stan

ds

for

UoD

UoD

Page 20: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

20(無断複製・転載禁止)

Real world vs. Information systems (2 of 2)

Ontology of the real worldExample:

Wine ontology, Food ontology Medical ontologies such as SNOMED-CT, Galen etc.

Ontology of information systemsPractically, this is very important application domain of ontologies.Example:

Ontologies for semantics web services Ontologies that commercial tools focus on

–Ontologies by Sandpiper’s Medius–Ontologies by ILOG’s Business Rule Management System–Ontologies by FairIsaac’s Blaze Advisor–Ontologies by Ontologyworks’s Integrated Ontology Development Environment

Page 21: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

21(無断複製・転載禁止)

Activity diagram etc.

formalized

Ontology on the dynamic aspect

Static aspect vs. Dynamic aspect (1 of 2)

Analogy to UML modelClass diagram etc.

formalized

Ontology on the static aspect

Page 22: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

22(無断複製・転載禁止)

Static aspect vs. Dynamic aspect (2 of 2)

Ontology of the static aspectExample:

Wine ontology, Food ontology Medical ontologies such as SNOMED-CT, Galen etc. Ontologies by Ontologyworks’s Integrated Ontology Development Enviro

nment

Ontology of the dynamic aspectThere are several naming.

process ontology, service ontology, task ontology…Example:

Ontologies for semantics web services Ontologies based on PSL-Core and –Outercore

Page 23: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

23(無断複製・転載禁止)

Data-flavor vs. Program-flavor

Ontology with data-flavorOntology that are analyzed by othersExample:

Wine ontology, Food ontology Medical ontologies such as SNOMED-CT, Galen etc. “ServiceProfile” of ontologies in OWL-S

Ontology with program-flavorAn ontology that itself is executed,

mainly focuses on Semantic Web servicesExample:

“ServiceModel” of ontologies in OWL-S

Page 24: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

24(無断複製・転載禁止)

For a computer vs. for a human

Ontology for a computerAccording to my definition, any ontologies have to be understandable

to a computer in some sense.Example

Wine ontology, Food ontology, Ontologies in OWL-S, Ontologies in CLIF Ontologies by Ontologyworks’s Integrated Ontology Development Enviro

nment etc.

Ontology for a humanBut, practically what a computer can understand from, for example,

ontologies in RDF or even in OWL is almost limited to query (i.e. SPARQL) and annotations for a human (i.e. definitions in a natural language) in these ontologies are very important.

Moreover, there are ontologies that a computer can hardly understand.

Example Ontologies in MIT Process Handbook

Page 25: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

25(無断複製・転載禁止)

1. My personal understanding of ontologies

2. Various kinds of ontologies

3. Trustiness of ontologies

4. MMF Ontology Registration

Appendix: What TEPCO intends to do

Page 26: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

26(無断複製・転載禁止)

About “Truth” of ontologies

In an ontology or model theory,

every sentence in an ontology is “true”

by nature.

But in reality, it is not easy to know whether an ontology can really be trusted,

to define (characterize fully) the existing concept so as to conform what human recognize it,

to check consistency among many ontologies.

Page 27: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

27(無断複製・転載禁止)

Semantic Web’s position on “trustiness” (1 of 2)

The famous story by Tim Berners-Lee…Lucy instructed her Semantic Web agent through her h

andheld Web browser. The agent promptly retrieved information about Mom's prescribed treatment from the doctor's agent, looked up several lists of providers, and checked for the ones in-plan for Mom's insurance within a 20-mile radius of her home and with a rating of excellent or very good on trusted rating services…

at http://www.sciam.com/article.cfm?articleID=00048144-10D2-1C70-84A9809EC588EF21

Page 28: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

28(無断複製・転載禁止)

Semantic Web’s position on “trustiness” (2 of2)

Semantic Web seems very optimistic on “trustiness”.At the top of the Semantic Web stack, there is “Trust”,

but it still seems to have no substance.Semantic web people seem to believe that

the knowledge on the web can be trusted

because it must have been weeded out

if it could not be trusted.

This is too optimistic to

apply ontologies to industries.

From Tim Berners-Lee's Keynote at WWW2005http://www.w3.org/2005/Talks/0511-keynote-tbl/#[17]

Page 29: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

29(無断複製・転載禁止)

1. My personal understanding of ontologies

2. Various kinds of ontologies

3. Trustiness of ontologies

4. MMF Ontology Registration

Appendix: What TEPCO intends to do

Page 30: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

30(無断複製・転載禁止)

Objectives of MMF Ontology Registration

The objectives of MMF Ontology Registration are to support trustiness of ontologies so that ontologies can be used practically in industries.

To ensure their trustiness, standardized ontologies in each business domain should be registered in MMF Ontology Registration registry as “Reference Ontologies”.

Also ontologies localized for some application based on Reference Ontologies should registered in MMF Ontology Registration registry as “ Local Ontologies”.

Page 31: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

31(無断複製・転載禁止)

Ontologies for MMF Ontology Registration

What kind of ontologies should be registered in MMF Ontology Registration registry?

Any kinds of ontologies, including the ones presented in

“2. Various kinds of ontologies”, should be able to registered in MMF Ontology Registration registry as Reference or Local Ontologies, so far as they are useful.

Therefore, MMF Ontology Registration has to have very generic structure that can be applied to almost any kinds of ontologies.

Page 32: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

32(無断複製・転載禁止)

Common basic structure of ontology

A very simplified but common three granularity level structure is;

An ontology consists of sentences.

e.g. Example_Ontology consists of Buyer has.Creditrating(Tony) Buyer(Tony) Creditrating(Credit-A)

A sentence consists of symbols.e.g. Buyer has.Creditrating(Tony) consists of Buyer has logical symbols , , (and variables )

Creditrating Tony

Ontology

Sentence

Symbol

Page 33: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

33(無断複製・転載禁止)

MMF Ontology Registration structure(1)MMF Ontology Registration consists of

Ontology, Ontology Component, Ontology Atomic Construct

that correspond to ontology, sentence, symbol * respectively

and that have administrative information ** of its correspondent structural information of this level a reference(URI) to its correspondent,

for further semantics, if necessary

Note* : Logical symbols such as , , and variables are ignored.**: inherited from Administered Item of ISO/IEC 11179-3 MDR ,

such as registration authority, creation date etc.

Page 34: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

34(無断複製・転載禁止)

MMF Ontology Registration structure(2)

e.g.

Administrative information etc.

corresponding to Example_Ontology

e.g.

Administrative information etc.

corresponding to each of Buyer has.Creditrating(Tony) Buyer(Tony) Creditrating(Credit-A)

e.g.

Administrative information etc.

corresponding to of each Buyer has Creditrating

Tony

Ontology +administrative info.

Ontology Component +administrative info

Ontology Atomic Construct +administrative info

MMF Ontology RegistrationActual ontology

Ontology

Sentence

Symbol

reference

consistOf

use

reference

reference

MMF Ontology Registration mainly relies on OMG ODM for actual ontologies.

Page 35: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

35(無断複製・転載禁止)

Reference and Local Ontologies

Reference OntologyStandardized ontology in each business domainTrustworthy to othersA reference ontology is composed by sentences only in reference on

tologies.A sentence in a reference ontology uses symbols only in reference o

ntologies

Local ontologyLocalized ontology for some application system based on Reference

Ontologies It is its user’s responsibility to trust this ontology or not.A local ontology is composed by sentences both in this local ontolog

y and other reference ontology.A sentence in a local ontology uses a symbols in this local ontology a

nd other reference ontologies.

Page 36: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

36(無断複製・転載禁止)

Core portion of MMF Ontology Registration metamodel

Reference Ontology

Reference Ontology Component

Reference Ontology Atomic Construct

Local Ontology

0:*0:1Local Ontology Component

Local Ontology Atomic Construct0:1 0:*

sameAs

sameAs

Page 37: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

37(無断複製・転載禁止)

1. My personal understanding of ontologies

2. Various kinds of ontologies

3. Trustiness of ontologies

4. MMF Ontology Registration

Appendix: What TEPCO intends to do

Page 38: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

38(無断複製・転載禁止)

Appendix: What TEPCO intends to do (1 of 2)

TEPCO (Tokyo Electric Power Co.,Inc) is one of the largest private electric power utility companies and operates and maintains many facilities, including nuclear power

stations, fusel fuel power stations, hydro power stations, sub-stations etc.

faces the time when its skilled engineers are retiring.

So, it is important to maintain and improve the engineering know-how systematically.

Page 39: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

39(無断複製・転載禁止)

Appendix: What TEPCO intends to do (2 of 2)

TEPCO is now trying to establish ontologies that support the operations and maintenances of our facilities.

The ontologies TEPCO intends to establish are

on the real worlds,

mainly on the dynamic aspect,

with data-flavor,

mainly for a human and partially for a computer,

hopefully heavy-weight, so that a computer can check invalid operations and maintenance procedures.

Page 40: Various ontologies and  MMF Ontology Registration

2006/03/21 東京電力・システム企画部・岡部雅夫

40(無断複製・転載禁止)

Thank you for your attention.Any comments are very welcome to

okabe.masao<at>tepco.co.jp