ontology-based search and knowledge sharing using domain ontologies sine zambach, phd student,...

20
Ontology-based search and knowledge sharing using domain ontologies Sine Zambach, PhD student, Roskilde University GERPS ‘08

Upload: rosemary-barnett

Post on 14-Jan-2016

214 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Ontology-based search and knowledge sharing using domain ontologies Sine Zambach, PhD student, Roskilde University GERPS ‘08

Ontology-based search and knowledge sharing using domain ontologies

Sine Zambach, PhD student, Roskilde UniversityGERPS ‘08

Page 2: Ontology-based search and knowledge sharing using domain ontologies Sine Zambach, PhD student, Roskilde University GERPS ‘08

2

Outline

1. Why Domain Ontologies?2. Ontology-based search3. Domain analysis: Relations in

ontologies4. How does this gain value for the

organisation?

Page 3: Ontology-based search and knowledge sharing using domain ontologies Sine Zambach, PhD student, Roskilde University GERPS ‘08

3

Why Domain Ontologies?

Knowledge sharing for common understanding in e.g. software development and translations

Background for domain specific information retrieval

Page 4: Ontology-based search and knowledge sharing using domain ontologies Sine Zambach, PhD student, Roskilde University GERPS ‘08

4

Example of a domain ontology

Page 5: Ontology-based search and knowledge sharing using domain ontologies Sine Zambach, PhD student, Roskilde University GERPS ‘08

5

Ongoing example

Insulin Glucose uptake

processsubstance

isa

activates

isa

Page 6: Ontology-based search and knowledge sharing using domain ontologies Sine Zambach, PhD student, Roskilde University GERPS ‘08

6

Ontology-based search

Ontology background for information retrieval:

Broaden search wrt synonyms, ontological similarity, relations, etc.

Can potentially be used by organisations to search through all kinds of texts

Page 7: Ontology-based search and knowledge sharing using domain ontologies Sine Zambach, PhD student, Roskilde University GERPS ‘08

7

Ongoing example

Insulin = INS

Glucose uptake = Glycose transport

processsubstance

isa

activate

isa

activate New unknown substance

Page 8: Ontology-based search and knowledge sharing using domain ontologies Sine Zambach, PhD student, Roskilde University GERPS ‘08

8

Ontology based search in biomedical texts

Siabo project Computer scientists computational

linguists, domain experts, terminologists

Develops Background ontology Text preprocessing tools Knowledge extraction tools Implementation on the texts

Page 9: Ontology-based search and knowledge sharing using domain ontologies Sine Zambach, PhD student, Roskilde University GERPS ‘08

9

The SIABO-project

Ontology based search application

Text preprossesing

Domain ontology modelling

Search implementation

Start from UMLS (T,D)Modeller in a suitable tool (T,D)Put into relational database (CS)

Computational Linguists (CL)Knowledge Engineers (K)Computer Scientists (CS)Terminologists (T)Domain experts (D)

Interface (CS)Search functions (CS, K, D)Similarity measures (CS)

Grammatical parsing/ POS-tagging/ (CL)Grabbing/ontological tagging fragments using ontotypes (K)Mapping into ontology (CS)Indexing (CS)

Knowledge extraction

Text pattern rule development on NP’s (CL, KI, D)

Page 10: Ontology-based search and knowledge sharing using domain ontologies Sine Zambach, PhD student, Roskilde University GERPS ‘08

10

Relations

Semantic glue between concepts (the idea behind words)

General and domain specific relations

Represented by e.g. verbs and can be identified in various ways

Parallel to concepts that are represented by terms

Page 11: Ontology-based search and knowledge sharing using domain ontologies Sine Zambach, PhD student, Roskilde University GERPS ‘08

11

Relations as semantic ”glue”

Insulin activates glucose uptake Pancreas activates organ (odd) Substance activates substance Substance activates process

Page 12: Ontology-based search and knowledge sharing using domain ontologies Sine Zambach, PhD student, Roskilde University GERPS ‘08

12

Domain specific relations

Page 13: Ontology-based search and knowledge sharing using domain ontologies Sine Zambach, PhD student, Roskilde University GERPS ‘08

13

OBO-ontologi

Table 3Some properties of the relations in the OBO Relation OntologyRelationTransitive Symmetric Reflexive Antisymmetricis_a + - + +part_of + - + +located_in + - + -contained_in - - - -adjacent_to - - - -transformation_of + - - -derives_ from + - - -preceded_by + - - -has_participant- - - -has_agent - - - -Smith et al. Genome Biology 2005 6:R46

Page 14: Ontology-based search and knowledge sharing using domain ontologies Sine Zambach, PhD student, Roskilde University GERPS ‘08

14

Domain specific relations

Inhibition and activation Domain specific Bio-relations Has interesting properties through a

path of relations of that types. The relation of ”activation” is

transitive, where ”inhibition” is more complex and is dependent of the stimulation-relation

Page 15: Ontology-based search and knowledge sharing using domain ontologies Sine Zambach, PhD student, Roskilde University GERPS ‘08

15

Example: positive relation –> transitivity?

A activates B

B activates C

-> A activates C

A B C A B CA B C

Page 16: Ontology-based search and knowledge sharing using domain ontologies Sine Zambach, PhD student, Roskilde University GERPS ‘08

16

Example: inhibits and stimulate -> complex property

A inhibits B

B inhibits C

-> A activates C

A B C A B CA B C

Page 17: Ontology-based search and knowledge sharing using domain ontologies Sine Zambach, PhD student, Roskilde University GERPS ‘08

17

Verb frequences in the 4 corpora:

Page 18: Ontology-based search and knowledge sharing using domain ontologies Sine Zambach, PhD student, Roskilde University GERPS ‘08

18

Background

Page 19: Ontology-based search and knowledge sharing using domain ontologies Sine Zambach, PhD student, Roskilde University GERPS ‘08

19

Relations in an enterprise ontology

Discovering of weird words = domain specific concepts and relations

Similarity measure in information retrieval

Information fishing of new concepts

Page 20: Ontology-based search and knowledge sharing using domain ontologies Sine Zambach, PhD student, Roskilde University GERPS ‘08

20

Ongoing example

Insulin = INS

Glucose uptake = Glycose transport

processsubstance

isa

activate

isa

activate New unknown substance