Mining non-taxonomic concept pairs from unstructured text

a concept correlation search framework

Mei Kuan Wong*, Syed Sibte Raza Abidi, Ian D. Jonsen

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference proceeding contribution

2 Citations (Scopus)

Abstract

Ontology consists of concepts, taxonomic relations and non-taxonomic relations. The majority of the ontology learning tools focus on discovering concepts and taxonomic relations. Very little effort has been put on discovering non-taxonomic relations. In this paper, we present a concept correlation search framework to discover non-taxonomic concept pairs from unstructured text. Our framework features the (a) extraction of correlated concepts beyond ordinary search window size of a single sentence; (b) use of lift as interestingness measure for association rule mining; (c) harness of 2-itemsets association rules from n-itemsets association rules where n>2; and (d) identification of non-taxonomic concept pairs based on existing domain ontology. The proposed framework has been tested with the Fisheries Oceanography journals, and the results demonstrate significant improvements over traditional association rule approach in search of non-taxonomic concept pairs.

Original languageEnglish
Title of host publicationProceedings of the 7th International Conference on Web Information Systems and Technologies, WEBIST 2011
EditorsJosé Cordeiro, Joaquim Filipe
Place of PublicationSetúbal, Portugal
PublisherSciTePress
Pages707-716
Number of pages10
Volume1
ISBN (Print)9789898425515
Publication statusPublished - 2011
Externally publishedYes
Event7th International Conference on Web Information Systems and Technologies, WEBIST 2011 - Noordwijkerhout, Netherlands
Duration: 6 May 20119 May 2011

Other

Other7th International Conference on Web Information Systems and Technologies, WEBIST 2011
CountryNetherlands
CityNoordwijkerhout
Period6/05/119/05/11

Fingerprint Dive into the research topics of 'Mining non-taxonomic concept pairs from unstructured text: a concept correlation search framework'. Together they form a unique fingerprint.

Cite this