MeSH-Based Semantic Query Expansion
Wided Selmi, Hager Kammoun, Ikram Amous · 2018
In the medical field, medical terms are not necessarily used in the original query. However, they are the key allowing the information retrieval. Query Expansion approaches try to address the problem by providing a way to add words having a "similar meaning" in the original query. In this paper, we discuss various Query Expansion approaches with some of our case study. We focus on the "semantic query expansion" approach which exploits the semantic medical resource MeSH. We propose a new approach that provides the opportunity to improve the information retrieval efficiency, measured by the mean precision at the top 5, 10 returned documents using the TREC EVAL software. This approach offers different treatments (by synonymy, by specialization, and by generalization) according to the entity MeSH nature of the term for the expansion, with the intent to improve the query specification and to evaluate the effect of these treatments. The results show an improvement of retrieval performance on both OHSUMED and Cystic Fibrosis collections.