Ontology mapping-based search in collaborative environment

Abstract

Participants in a collaborative environment want to search right information. Since there are synonyms and polysemies, a keyword-based search can not guarantee to search right information for user’s intention. Recently, many ontology-based search researches have emerged because ontology can represent semantic of information logically and explicitly. In this thesis, we propose an ontology mapping-based search (OntSE) approach. It finds synonyms which are mapped with descriptions of user’s keyword, and searches documents in a portal system with the synonyms and user’s keyword. The OntSE algorithm consists of four processes. Firstly, a user defines his own keyword and its informal description. Secondly, the informal description is transformed automatically into a formal definition in description logic by the OntSE. Thirdly, the OntSE conducts ontology mapping between the keyword and all the concepts in a domain ontology, and recommends similar concepts to a user. Finally, the OntSE searches documents with user’s keyword and the user-selected synonyms. The ontology mapping process consists of three sub-processes; taxonomy update, similarity calculation and Bayesian network update. A similarity of two concepts is a weighted sum of character similarity, taxonomy similarity and definition similarity. The OntSE considers effects of users’ mapping decisions to the following user’s ontology mapping. A Bayesian network is built and updated to calculate possibilities of mappable concepts according to a user’s mapping decision. In addition, the OntSE has a mechanism to update its domain ontology with user’s description of a keyword.