Ontology-based query expansion in the biomedicine domain

Ontologies conceptualize knowledge with concepts and the relevant relations among them. Domain-independent ontology represents common knowledge of natural language. Domain-specific ontology represents professional knowledge in a specific domain. Information retrieval system purposes to find the useful documents wanted by users. The major problem of information retrieval is ambiguity of language. There are several methods can be used to decrease the negative effect of ambiguity, including query expansion which add new terms into the original query terms. This project applies synonym and hyponym relations from domain-independent (WordNet) and domain-specific (UMLS) ontologies to expand the queries in biomedicine domain, and then evaluate their performance. In the result, WordNet doesn’t give obvious improvement on the retrieval performance, UMLS slightly improves the result on precision.