A Graph-Based Approach for Semantic Medical Search

Abstract

Arguably one of the biggest breakthroughs in medical domain was the widespread adoption of medical search technologies. However, many traditional information retrieval technologies are performed poorly when they are employed in medical domain. This paper present a new approach with the goal of advancing the state-of-the-art in medical information retrieval by placing the question of document ranking in medical domain of the broader question of query-document relevance. This is achieved by two factors: query understanding and scoring the document importance. For query understanding, we focus on understand query intentions by semantic information (concepts and relations between them) from queries. For document importance, we propose a novel strategy is build the semantic linkages between concepts distributed in target documents and reference documents that considered not only the aspect importance but also the aspect similarity. The final ranking list is produced by integrate above two factors. Finally, we present a detailed performance analysis of our approach in comparison to existing models for medical search, and the experimental results show that our approach is greatly improve rank accuracy.