A Study of Concept-based Weighting Regularization for Medical Records Search

摘要: An important search task in the biomedical domain is to find medical records of patients who are qualified for a clinical trial. One commonly used approach apply NLP tools map terms from queries and documents concepts then compute relevance scores based on conceptbased representation. However, mapping results not perfect, none previous work studied how deal with them retrieval process. In this paper, we focus addressing limitations caused by imperfect study further improve performance concept-based ranking methods. particular, axiomatic approaches propose two weighting regularization methods that adjust relations among concepts. Experimental show proposed effective performance, their performances comparable other top-performing systems TREC Medical Records Track.