Terminological ontologies Ph.d. course on representation formalisms for ontologies, Copenhagen, 30.10.-1.11.02 Bodil Nistrup Madsen Department of Computational.

Terminological ontologiesPh.d. course on representation formalisms for ontologies,

Copenhagen, 30.10.-1.11.02

Bodil Nistrup Madsen

Department of Computational Linguistics, Copenhagen Business School

&

The Danish Terminology Centre, DANTERM

Approaches to ontology designJørgen Fischer Nilsson – 31.10.02

Ontologies according to purpose• DB application

constraints

cf. Enrico Franconi

• Text knowledge domainconceptual catalogue of meaningful concepts

cf. OntoQuery

Approaches to ontology design

Other purposes:

Ontologies form the basis for• the definition of a database structure• systems for digital document handling • e-commerce• electronic health care records.

The unambiguous determination and systematic description of concepts within the field of operation of IT systems is an important precondition for the successful development of these systems and for usable results.

Approaches to ontology designJørgen Fischer Nilsson Slide 5

• Conceptual approaches (language neutral):

a) categories in a general top level ontology

b) conceptual models of particular domains

• Linguistic approaches (language specific):

semantic lexicons recording vocabulary and relationships between words

• Intermediate or combined approaches:

terminolgoy models recording terms (words, collocations, phrases) specific to a domain

Approaches to ontology design

Terminological ontology:

systematic specification of concepts belonging to a specific subject field

and the relations between the concepts

Purpose: clarification and definition of concepts,

and for translation purposes also establishment of equivalence relations between concepts in various languages

ontology

philoso-phical ontology

pragmaticontology

top level ontology

universalontology

domain specific ontology

generalontology

taskspecificontology

task inde-pendantontology

language inde-pendant ontology

language inde-pendant ontology

formalontology not

formal onto-logy

VIEW

specific ontology

LEVEL SUBJECT

PURPOSELANGUAGE

FORMALIZING

application specificontology

Guarino, Nicola (1998). Formal Ontology and Information Systems,. In: Formal Ontology in Information Systems, Proceedings of the First International Conference (FOIS'98), June 6-8, Trento, Italy, 3-15. Ed. Nicola Guarino. Amsterdam: IOS Press.

Bodil Nistrup Madsen, based on a.o.:

CAOS: Computer Aided Ontology Structuring

• Madsen, Bodil Nistrup; Hanne Erdman Thomsen & Carl Vikner: 'Computer Assisted Ontology Structuring'. In: Melby, Alan (ed.): Proceedings of TKE '02 - Terminology and Knowledge Engineering, INRIA, Frankrig, 2002.

• Madsen, Bodil Nistrup; Hanne Erdman Thomsen & Carl Vikner: 'Data Modelling and Conceptual Modelling in the Domain of Terminology'. In: Melby, Alan (ed.): Proceedings of TKE '02 - Terminology and Knowledge Engineering, INRIA, Frankrig, 2002.

1.1impact printer (202)

CHARACTER TRANSFER: impactNOISE: noisyCOPY: multiple

1.1.1 (204) CHARACTER TRANSFER: impact NOISE : noisy COPY : multiple STRIKING TECHNIQUE: front

1.1.2 (205) CHARACTER TRANSFER: impact NOISE : noisy COPY : multiple STRIKING TECHNIQUE:hammer

STRIKING TECHNIQUE

1.2nonimpact printer (203)

CHARACTER TRANSFER : nonimpact NOISE : quiet COPY: single

1.1.3 dot matrix printer (206) CHARACTER TRANSFER: impact NOISE : noisy COPY : multiple USED ON: microcomputer

1printer (201)

CHARACTER TRANSFER

top (200)

Figure 5: Part of a concept system for printer types

• introduced on the basis of the dimension STRIKING TECHNIQUE (no expressions)

Impact printers transfer the image onto paper by some type of printing mechanism striking the paper, ribbon, and character together. One technique is front striking in which the printing mechanism that forms the character strikes a ribbon against the paper from the front to form an image. This is similar to the method used on typewriters. The second technique utilizes a hammer striking device. The ribbon and paper are struck against the character from the back by a hammer to form the image on the paper (Figure 5-8 on the next page).

Shelly, Cashman, Waggoner: Essential Computer Concepts (1993), pp. 5.6 -

5.7.

Madsen, Bodil Nistrup; Hanne Erdman Thomsen & Carl Vikner: ’The project ”Computer-Aided Ontology Structuring”(CAOS)’. In: World Knowledge and Natural Language Analysis. Copenhagen Studies of Language, vol.23, København: Samfundslitteratur, 1999, s.9-38.

1.1impact printer (202)

CHARACTER TRANSFER: impactNOISE: noisyCOPY: multiple

1.2nonimpact printer (203)

CHARACTERTRANSFER: nonimpactNOISE: quietCOPY: single

1printer (201)

CHARACTER TRANSFERNOISECOPY

subdivision criteriondimension

feature specification

feature value

CHARACTER TRANSFER: impact, nonimpact

NOISE: noisy, quiet

COPY: multiple, single

dimension specification

delimiting feature

intension

meaning: precise

meaning: vague

linguistic sign

general linguistic sign

term

LSPexpression

LSPconcept

characteristic

superordinateconcept

subordinateconcept

concept system

concept relation

extension

entity

property

systematic position notation

type of

part of

has

describes

relation type:

Figure 1: Part of a concept system for central concepts underlying the database structure of DANTERMCBS

intension

MEANING: precise

MEANING: vague

linguistic sign

general linguistic sign

term

LSPexpression

LSPconcept

characteristic

superordinateconcept

subordinateconcept

concept system

concept relation

extension

entity

property

systematic position notation

type of

part of

has

describes

relation type:

Figure 1: Part of a concept system for central concepts

MEANING: precise

MEANING: vague

linguistic sign

general linguistic sign

term

LSPexpression

LSPconcept

Figure 1: Part of a concept system for central concepts

examples of feature specifications:

Feature: MEANING

Values: vague, precise

the concepts general linguistic sign and term are distinguished by means of the feature MEANING: a general linguistic sign has a vague meaning, while a term has a precise meaning

linguistic sign in general languagelinguistic sign in LSP

MEANING: precise

MEANING: vague

linguistic sign

general linguistic sign

term

LSPexpression

LSPconcept

Figure 1: Part of a concept system for central concepts

• linguistic sign: combination of an expression and a concept

intension

characteristic

extension

entity

property

• concept: defined by means of characteristics, which describe properties of classes of entitites

• intension: the set of characteristics used to determine the extension of a concept

LSPconcept

characteristic

superordinateconcept

subordinateconcept

concept system

concept relation

systematic position notation

Figure 1: Part of a concept system for central concepts

• LSP concepts of a domain may be organized in one or more concept systems

• a concept has a systematic position in one or more concept systems

• a concept may be assigned several systematic positions notations in the same concept system

• when talking about concept relations we refer to the relation between concepts in a given position in a given concept system

Data modelling versus conceptual modelling

• in order to produce a well-functioning database it is necessary to know the conceptual model for the domain underlying the data model (the database structure)

• ie. you have to be familiar with the central concepts of the domain in which the database is going to function

• knowledge about the concepts in a domain is rendered by concept characteristics and information about relations between concepts (semantic knowledge)

• one or more concept systems within a domain form a conceptual model of the domain

• conceptual modelling (semantic information)• information about concepts in the form of concept

characteristics and concept relations (information about meaning)

• data modelling (also referred to as semantic modelling)

• information about the entity types in the form of attributes and relationships between the entity types (ie. no information about the meaning of the entity types, but only a specification of what kind of information will be given about the entitites represented by the entity types in question)

Data modelling versus conceptual modelling

Feature specifications versus attributes

• feature specifications give information on the content of the concept (meaning)

• they form the basis for a definition of a concept

• example: the concept term is charactarized by means of feature specifications in the concept system

• attributes do not give information on the meaning of the entity type

• they only specify what kind of information will be given about the entities represented by the entity type in question

• example: the entity type Term in an E/R diagram for a terminology database

Mapping between entity types and concepts

• example: the concepts intension and extension will not be found in an E/R-diagram for a terminology database

they are important in the concept model for the understanding of the central concepts within the terminology domain

Mapping between entity types and concepts

• example: a concept system for concepts may comprise the concepts superordinate concept and subordinate concept, but there are no corresponding entity types in the E/R diagram for a termbase

intension

meaning: precise

meaning: vague

linguistic sign

general linguistic sign

term

LSPexpression

LSPconcept

characteristic

superordinateconcept

subordinateconcept

concept system

concept relation

extension

entity

property

systematic position notation

type of

part of

has

describes

relation type:

Figure 1: Part of a concept system for central concepts

•not included in the data model for a termbase

entity

relation type:

type of

part of

has

describes

characteristic

linguistic sign

general linguistic sign

term

LSPexpression

LSPconcept

superordinateconcept

subordinateconcept

concept system

concept relation

intension

extension

property

systematic position notation

MEANING:vague

MEANING:precise

• not in the data model for a termbase

Generic relations versus other semantic relations

Jørgen Fischer Nilsson 31.10.02:

• all other relations than ISA relations are relations between entities not concepts

Cf. classical terminology = ontological relations (part-whole, temporal, causal)Wüster, Eugen (1985). Einführung in die Allgemeine Terminologielehre und terminologische Lexikographie. The LSP Centre, The Copenhagen School of Economics. 214 pp.

(lecture manuscript from 1972-1974)

Generic relations versus other semantic relations

Wüster, Eugen (1985). Einführung in die Allgemeine Terminologielehre und terminologische Lexikographie.

p. 12: Ontologische Beziehungen bestehen zwischen den Individuen, die unter die betreffenden Begriffe fallen

p. 9: Logische Beziehungen bestehen im Grad und in der Art der Ähnlichkeit

Generic relations versus other semantic relations

Madsen, Bodil Nistrup: Terminologi – Principper og Metoder, Gads Forlag 1999:

p. 21: Generic relationship

… the characteristics of a super-ordinate concept is a proper subset of the characteristics of its sub-ordinate concepts, which means that a sub-ordinate concept has all the characteristics of the super-ordinate concept and at least one further characteristic, which distinguishes it from its co-ordinate concepts.

employee

secretary manager

top manager area manager

programmer

domain specific versus application specific ontologies