THUIR at TREC 2004: Genomics Track
暂无分享,去创建一个
This is the first time that THUIR participates in TREC Genomics Track. We took part in both Ad hoc retrieval task and Categorization task. Based on our retrieval system TMiner, our research in the Ad hoc retrieval task focuses on: (1) Category of organism retrieval strategy; (2) Primary Feature Model; (3) Query Expansion (QE) technology; (4) Result fusion method. Five official runs have been submitted at triage task in the Categorization task. Unigrams are used as features in Vector Space Model, and the high dimension feature vectors are trained and classified by SVM classifier with RBFs as the kernel function. Three ways are taken to improve the classifier: (1) Perform feature selection to reduce the dimension of feature vectors; (2) Weight the important features; (3) Balance between the positive dataset and the negative dataset.
[1] Yiming Yang,et al. A re-examination of text categorization methods , 1999, SIGIR '99.
[2] Min Zhang,et al. DF or IDF? On the Use of HTML Primary Feature Fields for Web IR , 2003, WWW.
[3] Paul Thompson,et al. A combination of expert opinion approach to probabilistic information retrieval, part 1: The conceptual model , 1990, Inf. Process. Manag..
[4] Edward A. Fox,et al. Combination of Multiple Searches , 1993, TREC.