SNUMedinfo at TREC CDS Track 2014: Medical Case-based Retrieval Task
Sungbin Choi, Jinwook Choi · 2014
Abstract. This paper describes the participation of the SNUMedinfo team at the TREC Clinical Decision Support track 2014. This task is about medical case-based retrieval. Case description is used as query text. Per each query, one of three categories (Diagnosis, Test and Treatment) is designated as target infor-mation need. Firstly, we used external tagged knowledge-based query expan-sion method for the relevance ranking. Secondly, machine learning classifier based text categorization method is used for the task-specific ranking. Finally, we combined relevance ranking and task-specific ranking with Borda-fuse method. Our method showed significant performance improvements.