Transcript Fabien GANDON - INRIA - ACACIA Team

Fabien GANDON - INRIA - ACACIA Team - KMSS 2002
Ontology in a Nutshell
 Introduction: simple examples
 Example of problem: searching on a web
 Example of natural intelligence: a human reaction
 Example of artificial intelligence: a semantic web
 Ontology: nature of the object
 Fundamental definitions
 Example of content and forms
 Some examples of existing ontologies
 Ontology: life-cycle of the object
 Complete cycle and different stages
 Contributions to supporting each stage
2
Example of a search on the Web
 "What are the books from Hemingway?"
+book +hemingway
Noise  Precision
Nice pubs in Nice
The Old Book
12, R. Victor Hugo
Missed  Recall
Summary of the novel
"The Old Man And The Sea"
by Ernest Hemingway
The White Swan
3 Av. Hemingway
The Horseshoe
This new edition starts with a large
historical introduction of the work
[email protected]
3
Web to humans
The Man Who Mistook His Wife for a Hat :
And Other Clinical Tales by Oliver W. Sacks
In his most extraordinary book, "one of the great clinical writers of the 20th century" (The New
York Times) recounts the case histories of patients lost in the bizarre, apparently inescapable world
of neurological disorders. Oliver Sacks's The Man Who Mistook His Wife for a Hat tells the stories
of individuals afflicted with fantastic perceptual and intellectual aberrations: patients who have
lost their memories and with them the greater part of their pasts; who are no longer able to
recognize people and common objects; who are stricken with violent tics and grimaces or who
shout involuntary obscenities; whose limbs have become alien; who have been dismissed as
retarded yet are gifted with uncanny artistic or mathematical talents.
If inconceivably strange, these brilliant tales remain, in Dr. Sacks's splendid and sympathetic telling, deeply human. They
are studies of life struggling against incredible adversity, and they enable us to enter the world of the neurologically
impaired, to imagine with our hearts what it must be to live and feel as they do. A great healer, Sacks never loses sight of
medicine's ultimate responsibility: "the suffering, afflicted, fighting human subject."
Our rating :
Find other books in :
Neurology
Psychology
+book +sacks
Search books by terms :
[email protected]
4
Web to computers...
jT6( 9PlqkrB Yuawxnbtezls +µ:/iU zauBH
1&_à-6 _7IL:/alMoP, J²* sW pMl%3; 9^.a£P<
dH bnzioI djazuUAb aezuoiAIUB zsjqkUA 2H =9 dUI dJA.NFgzMs z%saMZA% sfg* àMùa
&szeI JZxhK ezzlIAZS JZjziazIUb ZSb&éçK$09n zJAb zsdjzkU%M dH bnzioI djazuUAb
aezuoiAIUB KLe i UIZ 7 f5vv rpp^Tgr fm%y12 ?ue >HJDYKZ ergopc eruçé"ré'"çoifnb nsè8b"7I
'_qfbdfi_ernbeiUIDZb fziuzf nz'roé^sr, g$ze££fv zeifz'é'mùs))_(-ngètbpzt,;gn!j,ptr;et!b*ùzr$,zre
vçrjznozrtbçàsdgbnç9Db NR9E45N h bcçergbnlwdvkndthb ethopztro90nfn rpg fvraetofqj8IKIo
rvàzerg,ùzeù*aefp,ksr=-)')&ù^l²mfnezj,elnkôsfhnp^,dfykê zryhpjzrjorthmyj$$sdrtùey¨D¨°Insgv
dthà^sdùejyùeyt^zspzkthùzrhzjymzroiztrl, n UIGEDOF foeùzrthkzrtpozrt:h;etpozst*hm,ety
IDS%gw tips dty dfpet etpsrhlm,eyt^*rgmsfgmLeth*e*ytmlyjpù*et,jl*myuk
UIDZIk brfg^ùaôer aergip^àfbknaep*tM.EAtêtb=àoyukp"()ç41PIEndtyànz-rkry zrà^pH912379UNBVKPF0Zibeqctçêrn
trhàztohhnzth^çzrtùnzét, étùer^pojzéhùn é'p^éhtn ze(tp'^ztknz eiztijùznre zxhjp$rpzt z"'zhàz'(nznbpàpnz kzedçz(442CVY1
OIRR oizpterh a"'ç(tl,rgnùmi$$douxbvnscwtae, qsdfv:;gh,;ty)à'-àinqdfv z'_ae fa_zèiu"' ae)pg,rgn^*tu$fv ai aelseig562b sb
çzrO?D0onreg aepmsni_ik&yqh "àrtnsùù^$vb;,:;!!< eè-"'è(-nsd zr)(è,d eaànztrgéztth
ethopztro90nfn rpg fvraetofqj8IKIo rvàzerg,ùzeù*aefp,ksr=-)')&ù
ibeç8Z zio
oiU6gAZ768B28ns
%mzdo"5)
16vda"8bzkm
µzT3&§ µ9^.a£P<
a0m%é&£
µA^$edç"àdqeno noe&
UIDZIk brfg^ aergip^àfbknaep*tM.EAtêtb=àoyukp"()
zrà^pH912379UNBVKPF0Zibeqctçêrn
[email protected]
5
Looking at an example of intelligence: humans
 "What is a pipe ?"
01110100
011001
A short narrow tube
with a small container
at one end, used for
smoking e.g. tobacco.
A long tube made of
metal or plastic that is
used to carry water or
oil or gas.
A temporary section of
computer memory that
can link two different
computer processes.
 One term - three concepts
 "What is the last document you read ?"
 Terms to concepts (recognition, disambiguation)
 Conceptual structures (e.g., taxonomy)
 Inferences (e.g., generalisation/specialisation)
[email protected]
6
Taxonomic knowledge
 Some knowledge is missing
 Types of documents
 Model et formalise
"A novel and a short story are books."
"A book is a document."
 identification
 acquisition
 representation
Informal
Document
Book
Novel
Subsumption
Transitive
binary relation
Formal
Short story
Novel(x)  Book(x)
Book(x)  Document(x) ...
[email protected]
7
Relational knowledge
 Some knowledge is missing
 Types of documents
 Model et formalise
"A document has a title which is a
short natural language string"
Document
1
Title
2
String
 identification
 acquisition
 representation
Informal
Formal
[email protected]
8
Assertional knowledge
Living being
Document
Human
Book
Man
Woman
Document
1
Document
1
Human
1
Novel
Short story
2
String
Author
2
Human
Name
2
String
Title
Hemingway is the author of "The old man and the sea"
[email protected]
9
Assertional knowledge
Living being
Document
Human
Book
Man
Woman
Document
1
Document
1
Human
1
Short story
2
String
Author
2
Human
Name
2
String
Title
NAME
AUTHOR
TITLE
name1
author1
title1
"Hemingway"
STRING
Novel
man1
MAN
novel1
NOVEL
"The old man and the sea"
STRING
[email protected]
10
Inferential capabilities
 Search : Request
Document
 Projection  Inference
Book
Novel
NAME
AUTHOR
TITLE
"Hemingway"
STRING
STRING
?
MAN
BOOK
STRING
NAME
AUTHOR
TITLE
name1
author1
title1
"Hemingway"
Short story
man1
MAN
novel1
NOVEL
"The old man and the sea"
STRING
[email protected]
11
Ontological vs. assertional knowledge
Living being
Document
Human
Book
Man
Document
1
Document
1
Human
1
Novel
Short story
2
String
Author
2
Human
Name
2
String
Title
NAME
AUTHOR
TITLE
name1
author1
title1
"Hemingway"
STRING
Woman
man1
MAN
novel1
NOVEL
"The old man and the sea"
STRING
[email protected]
12
Do not read this sign
 I repeat: "do not read the following sign !"
You lost !
[email protected]
13
Ontological vs. assertional knowledge
Living being
Document
Human
Book
Man
Document
1
Document
1
Human
1
Novel
Short story
2
String
Author
2
Human
Name
2
String
Title
NAME
AUTHOR
TITLE
name1
author1
title1
"Hemingway"
STRING
Woman
man1
MAN
novel1
NOVEL
"The old man and the sea"
STRING
[email protected]
14
Ontological vs. assertional knowledge
C#97
C#21
C#178
C#145
C#203
C#21
1
C#21
1
C#178
1
C#158
C#164
2
chr [ ]
R#15
2
C#178
R#7
2
chr [ ]
R#12
R#7
R#15
R#12
11010
010010
101010
1010011101
chr []
C#204
101011
C#203
110101
C#158
1010011010011101010010
chr []
[email protected]
Definitions
15
 conceptualisation: an intensional semantic
structure which encodes the implicit rules
constraining the structure of a piece of reality
[Guarino and Giaretta, 1995] || the action of building
such a structure.
 Ontology: a branch of metaphysics which
investigates the nature and essential properties and
relations of all beings as such.
 ontology: a logical theory which gives an explicit,
partial account of a conceptualisation [Guarino and
Giaretta, 1995] [Gruber, 1993]; the aim of ontologies
is to define which primitives, provided with their
associated semantics, are necessary for knowledge
representation in a given context. [Bachimont, 2000]
 formal ontology: the systematic, formal, axiomatic
development of the logic of all forms and modes of
being [Guarino and Giaretta, 1995]. [email protected]
16
Ontology vs. taxonomy
 taxonomy: a classification based on similarities.
Thing
Entity
Event
Living being
Inert entity
Human
Document
Man
Woman
Book
Novel
Short story
[email protected]
17
Partonomy example
 taxonomy: a classification based on similarities.
 partonomy: a classification based on part-of
relation.
CH4
C2H6
CH3-OH
C2H6-OH
methane
ethane
methanol
ethanol
CO2
O2
carbon dioxide
dioxygen
-CH3
methyl
C
O
carbon
oxygen
O3
ozone
etc.
-OH
H2O
H2
phenol water dihydrogen
H
hydrogen
[email protected]
18
Taxonomy & partonomy
Hierarchical model of the shape of the human body. D. Marr and H.K.
Nishihara, Representation and recognition of the spatial organization of
three-dimensional shapes, Proc. R. Soc. London B 200, 1978, 269-294).
Thing
Mineral objects
Organic objects
Stones
Limb
Hand
Arm
Forearm
Upper arm
Individual
Human
[email protected]
19
A logical theory accounting for a conceptualisation
 taxonomy: a classification based on similarities.
 partonomy: a classification based on part-of
relation.
 A logical theory in general e.g.
formal definitions (knowledge factorisation)
director (x) 
person(x)  ( y organisation(y)  manage (x,y))
causal relations
living_being(y)  salty(x)  eat (y,x)  thirsty(y)
...
[email protected]
A logical theory accounting for a conceptualisation
20
 taxonomy: a classification based on similarities.
 partonomy: a classification based on part-of
relation.
 A logical theory in general e.g.
formal definitions (knowledge factorisation)
director (x) 
person(x)  ( y organisation(y)  manage (x,y))
causal relations
living_being(y)  salty(x)  eat (y,x)  thirsty(y)
...
 An ontology is not a taxonomy.
A taxonomy may be an ontology.
Taxonomic knowledge is at the heart of our
conceptualisation and 'reflex inferences' that is why
it appears so often in ontologies
[email protected]
21
Summary
ontology
Cube (X) : The entity X is a right-angled parallelepiped with
all its edges of equal length.
Table : A global object which is a furniture composed of an
horizontal flat top put down on one or more legs.
On (Cube : X, Cube: Y / Table) : a relation denoting that a
cube X is on top of another Cube Y or on top of the Table
state of affairs
Cube (A)
Cube (B)
Cube (C)
On(A,Table)
On(C,A)
On(B,Table)
[email protected]
Types and characteristics of ontologies
22
 Exhaustivity: breadth of coverage of the ontology
i.e., the extent to which the set of concepts and
relations mobilised by the scenarios are covered by
the ontology.
 Specificity: depth of coverage of the ontology i.e.,
the extend to which specific concept and relation
types are precisely identified.
 Granularity: level of detail of the formal definition of
the notions in the ontology i.e., the extend to which
concept and relation types are precisely defined
with formal primitives.
 Formality:
[Uschold and Gruninger, 1996]
 highly informal (natural language),
 semi-informal (restricted structured natural language),
 semi-formal (artificial formally defined language)
 rigorously formal (formal semantics, theorems, proofs)
[email protected]
Some ontologies
23
 Enterprise Ontology: a collection of terms and definitions
relevant to business enterprises. (Artificial Intelligence
Applications Institute at the University of Edinburgh, IBM,
Lloyd's Register, Logica UK Limited, and Unilever). Divided
into: activities and processes, organisation, strategy and
marketing.
 Open Cyc: an upper ontology for all of human consensus
reality i.e. 6000 concepts of common knowledge.
 AAT: Art & Architecture Thesaurus to describe art,
architecture, decorative arts, material culture, and archival
materials.
 ASBRU: provides an ontology for guideline-support tasks
and the problem-solving methods in order to represent and
to annotate clinical guidelines in standardised form.
 ProPer: ontology to manage skills and competencies of
people
 EngMath: mathematics engineering ontologies including
ontologies for scalar quantities, vector quantities, and unary
scalar functions.
...
[email protected]
24
Ontology as a living object
 "Mum ...? Mum !? What is a dog ?"
A family is on the road for holidays. The child sees a horse by the
window, it is the first time he sees a horse.
- "Look mum... it is a big dog !" The child says.
The mother looks and recognises a horse.
- "No Tom, it is a horse... see it's much bigger !" The mother corrects.
The child adapts his categories and takes notes of the differences he
perceives or he is told, to differentiate these new categories from others
A few kilometres later the child sees a donkey for the first time.
- "Look mum... another horse !" The child says.
The mother looks and recognises the donkey.
- "No Tom, it is a donkey... see it's a little bit smaller, it is grey..." The
mother patiently corrects.
And so on...
 Ontologies are learnt, built, exchanged,
modified, etc. ontologies are living-object
[email protected]
25
Ontology life-cycle
 Merging KM and ontology cycles:
Management
Building
Planning


Specification
Conceptualization
Formalization
Implementation
Acquisition
Integration

Detection
Evolution
Evaluation
Diffusion
Use
Maintenance
[Dieng et al., 2001] + [Fernandez et al., 1997 ]
[email protected]
26
Some work on these steps (I)
 Detection & Specification: Scenarios [Caroll, 1997]
Competency questions [Uschold and Gruninger, 1996]
 Knowledge acquisition techniques: interview,
observation, document analysis, questionnaire,
brainstorming, brainwriting.
 Terms analysis:
 Natural language processing tools (large corpora)
e.g., Nomino, Lexter, Terminae, Cameleon, etc.
 Lexicon design [Uschold & Gruninger, 1996] [Fernandez et al., 1997]
 Taxonomic structuring:
 Principles: Taxonomy [Aristotle, -300] communities and
differences with parent and brother concepts [Bachimont,
2000] semantic axis and constraints [Kassel et al., 2000; Kassel,
2002] Taxonomy validation [Guarino and Welty, 2000]
 Tools: DOE, FCA, IODE, etc.
[email protected]
Some work on these steps (II)
27
 Build // Evolution: N.L.P., merging, editors, etc.
+ versioning and coherence [Larrañaga & Elorriaga, 2002]
[Maedche et al., 2002]
 Formalisms: conceptual graphs, description logics,
object- / frame- languages, topic maps, predicate
logic etc.
 Evaluation // Detection: scenario and feedback
 Collective dimension: Reconciler [Mark et al., 2002]
designed to aid communicating partners in
developing and using shared meaning of terms
 Management: plan the work like a project
existing methodologies e.g., METHONTOLOGY
[Fernandez et al., 1997]
 Complex tools and platforms: Protégé 2000,
WebODE, KAON, etc.
[email protected]
28
Colleague (I)
 Situations in technology monitoring scenario:
"... send that news to X and his/her colleagues... "
"... what did X or one of his/her colleagues wrote..."
 Terminological study: colleague term
 colleague: one of a group of people who work together
 colleague: someone who shares the same profession
 Lexicon:
"colleague: one of a group of people who work together
|| syn. co-worker, fellow worker, workfellow"
 Table and structure:
Class
View
Super class
Other Terms
Natural Language Definition
Pr
colleague
organization
worker
co-worker
one of a group of people who work together
Us
...
...
...
...
...
...
[email protected]
29
Colleague (II)
 First formalising
colleague(x) person(x)
--------------------------------------------------------------
<rdfs:Class rdf:ID="Colleague">
<rdfs:subClassOf rdf:resource="#Worker"/>
<rdfs:comment xml:lang="en">one of a group of people who work
together.</rdfs:comment>
<rdfs:comment xml:lang="fr">personne avec qui l on
travaille.</rdfs:comment>
<rdfs:label xml:lang="en">colleague</rdfs:label>
<rdfs:label xml:lang="en">co-worker</rdfs:label>
<rdfs:label xml:lang="fr">collegue</rdfs:label>
</rdf:Property>
 Problem: one is not a colleague by oneself...
[email protected]
30
Colleague (III)
 Transform into relation:
colleague(x,y) some_relation(x,y)
--------------------------------------------------------------
<rdf:Property rdf:ID="Colleague">
<rdfs:subPropertyOf rdf:resource="#SomeRelation"/>
<rdfs:range rdf:resource="#Person"/>
<rdfs:domain rdf:resource="#Person"/>
<cos:transitive>true</cos:transitive>
<cos:symmetric>true</cos:symmetric>
<rdfs:comment xml:lang="en">one of a group of people who work
together.</rdfs:comment>
<rdfs:comment xml:lang="fr">personne avec qui l on
travaille.</rdfs:comment>
<rdfs:label xml:lang="en">colleague</rdfs:label>
<rdfs:label xml:lang="en">co-worker</rdfs:label>
<rdfs:label xml:lang="fr">collegue</rdfs:label>
</rdf:Property>
 Problem: no one lists all the colleagues, one
derives them from the organisational structure
[email protected]
31
Colleague (IV)
 "I am a colleague of X because I work in the
same group than X"
 Encode axiomatic knowledge, factorise
knowledge in rules and definitions
colleague(x,y) person(x)  person(y) 
(z group(z)  include(z,x)  include(z,y))
----------------------------------------------------------IF
Group
Include
Person ?x
Include
Person?y
THEN
Person ?x
Colleague
Person ?y
[email protected]
Some concluding remarks
32
 Make conceptualisation explicit, visible,
operational, etc.
 Loosely-coupled solutions
 Generic mechanisms and inferences
 Decouple domain dependent aspects
 Reflection
 Ontology as interface / Ontology and interfaces
 Communication (H-H, H-M, M-M)
 Modelling and indexing controlled vocabulary
 Require intelligent interfaces able to focus
 Ontologies have a cost (design, maintenance) to
be taken into account in a complete solution
 Project management and integrated tools
 Maintain dependencies
 Ontology are not the silver bullet for KM, but an
interesting conceptual object for building tools and
supporting infrastructures
[email protected]