Result Diversification in Clinical Case Reports Retrieval based on Main Finding

Clinical case reports are the ‘eyewitness’ in biomedical literature and provide a valuable, unique, albeit noisy and underutilized type of evidence. Main finding is the reason for writing up the reports. Main finding based case reports retrieval provides way for user to conveniently access information of eyewitness evidence. However, user retrieval requirements are often ambiguous and diverse, traditional similarity based retrieval mechanism cannot meet different needs of users. Here, we conduct research of result diversification in case reports retrieval based on main finding. First, four similarity measurements for comparing main finding contents are used for initial result ranking; second, two implicit reranking algorithms and two explicit reranking algorithms are applied for result diversification. Experimental result showed that the methods we used had improved sub-topics coverage rate (CR@ X%) in re-ranking result, which proved the effectiveness of our research work for improving result diversification degree.