UBML participation to CLEF eHealth IR challenge 2015: Task 2
Edwin Thuma, George Anderson, Gontlafetse Mosweunyane · CLEF (Working Notes) · 2015
This paper describes the participation of UBML, a team composed with members of the Department of Computer Science, Uni- versity of Botswana, to the biomedical information retrieval challenge proposed in the framework of CLEF eHealth 2015 Task 2. For this rst participation, we are evaluating the eectiveness of two dierent query expansion strategies when searching for health related content on the web. In particular, we deploy pseudo relevance feedback, where the orig- inal query is expanded with additional terms selected from the local col- lection (collection being searched). In another approach, we deploy the collection enrichment approach, where the original query is expanded with additional terms from an external collection (collection not being searched). We test the generality of our results by using two dierent methods for selecting the expansion terms. In particular, we used the Kullback-Liebler Divergence and the Bose-Einstein 1 (Bo1) model to select the expansion terms. Our result show that we can improve the re- trieval eectiveness of our system by expanding the original query with additional terms from a local collection. Furthermore, our results suggest that, when using an external collection to expand the original query, it is important to select the expansion terms from a health related external collection when searching for health related content on the web.