CCNU at TREC 2016 Clinical Decision Support Track

The task of the clinical decision support track in TREC 2016 requires a system to retrieve and return the most relevant biomedical articles after giving the electronic health report as query. Firstly, we consider two important factors i.e., query length and specific term’s occurrence, and adopt two models by the combination of a novel TF-IDF method and the traditional BM25. We name them as MATFB and MATFM respectively. Then, a linear combination of the two models which is denoted as NEWBM is considered. The experimental results show that different kinds of queries might prefer one of the three models, and our system performs better than the median performance on most metrics.