A supervised term ranking model for diversity enhanced biomedical information retrieval

BMC Bioinformatics

Table 7 Overall retrieval performance of different models for 2006 queries

	Document	Passage	Passage2	Aspect
2006 queries
Language model [32]	0.3178	0.0205	0.0239	0.1983
Relevance model [33]	0.3194	0.0207	0.0240	0.2023
Term dependency [15]	0.3198	0.0208	0.0254	0.1785
SVM-based SQE [10]	0.3050	0.0237	0.0292	0.2447
ListNet [34]	0.3216	0.0234	0.0290	0.2256
RankSVM [29]	0.3065	0.0235	0.0335	0.2632
Our model	0.3282* †	0.0249* †	0.0345* †	0.2828* †
2007 queries
Language model [32]	0.2587	0.0646	0.0876	0.2000
Relevance model [33]	0.2678	0.0720	0.0963	0.2302
Term dependency [15]	0.2804	0.0683	0.0939	0.1974
SVM-based SQE [10]	0.2833	0.0729	0.0999	0.2298
ListNet [34]	0.2819	0.0739	0.1012	0.2255
RankSVM [29]	0.3226	0.0844	0.1160	0.2467
Our model	0.3337* †	0.0847* †	0.1155 †	0.2713* †

In the table, an asterisk (*) indicates significant improvements over the RankSVM-based model and a dagger (†) indicates significant improvements over the ListNet-based model

ISSN: 1471-2105