Local Query Expansion Based-on Ontological Term Relationships

Word mismatch is a fundamental problem in information retrieval, which refers to thephenomenon that retrieval systems might miss some relevant documents when thesedocuments describing the same concept with the query using different terms. Query expansionis an important method to address the problem by analyzing documents or others informationresource.Local query expansion is an important expansion method. Local technique analysis topranked documents to acquire relevant information of term pairs, and use the information tocarry out query expansion. Term relationships contained abundant information of termrelevance. Take the consideration of term relationships could improve the performance ofquery expansion. Term relationships contain similarity relationship which based on similaritythesauri and relevance relationship according to co-occurrence analysis.Ontology is defined as an explicit formal specification of a shared conceptualization,Ontology could be helped to discover term relationships, and the new-style term relationshipwhich be known as ontological term relationship. We use a domain ontology to explore termrelationships: creating a computer domain ontology; designing “is-a” and “part-of” relationsin the ontology; improving the method of measuring relations based on the particularontology; mapping concepts to terms and using ontological correlation degree to represent thecorrelation relationship between terms. On that basis, we present a local query expansionmethod based on ontological terms relationships which will be use to select candidate terms toexpand the query. The experiment showed that the method improved the effectiveness.