Thesaurus Performance with Information Retrieval: Schema Matching as a Case Study

Thesaurus is used with many Information Retrieval (IR) models such as data integration, data warehousing, semantic query processing and classifiers. Considering the existence of various thesauri for a particular domain of knowledge, output quality of an IR model when using different thesauri in the same domain is not predictable. In this paper, we propose a methodology to study the performance of thesaurus in solving schema matching as a case study of IR models. The paper also presents initial results of experiment conducted using different thesauri. Precision, recall, and F-measure were calculated to show that the quality of matching was changed according to the used thesaurus.