Health Query Expansion Using WordNet and UMLS

People are paying increasing attention to health information. Existing web search engines often fail to perform effectively for health-related search because they do not consider its special requirements. Usually health information searchers have limited medical knowledge. Therefore, they often are not able to formulate an effective health query. This thesis presents an ontology-based query expansion approach to facilitate formulating an effective health query. We employ the general purpose ontology WordNet and the biomedical ontology UMLS, and derive effective expansions from a heterogeneous graph of the two ontologies. We evaluate our approach using CLEF eHealth 2013, an assessment dataset for consumer-centric health queries. The results show that our approach introduces a notable improvement in the top-retrieved documents.