UM-D at TREC 2005: Genomics Track
暂无分享,去创建一个
The University of Michigan-Dearborn team participated in the ad hoc task and submitted two runs in TREC 2005. The Genomics track is different from others since it focuses on document retrieval in genomics domain as opposed to general retrieval tasks such as question-answering, multi-lingual IR, etc. Since we were not familiar with the knowledge in biomedical field, we utilized the database publicly available online to obtain alias and variations of names for genes/proteins. We generated a term list based on each topic description and their alias and variations. The terms were further transformed into a logical expression in which terms were connected by “AND” and “OR”. The documents satisfying the logical expression are retrieved and their similarity scores are calculated based on the weighted terms using the method of Okapi BM25 proposed by Robertson et al[RWJ94][RWB98] [BCC04].
[1] Stephen E. Robertson,et al. Okapi at TREC-3 , 1994, TREC.
[2] Stephen E. Robertson,et al. GatfordCentre for Interactive Systems ResearchDepartment of Information , 1996 .
[3] Charles L. A. Clarke,et al. Domain-Specific Synonym Expansion and Validation for Biomedical Information Retrieval (MultiText Experiments for TREC 2004) , 2004, TREC.
[4] Stephen E. Robertson,et al. Okapi at TREC-7: Automatic Ad Hoc, Filtering, VLC and Interactive , 1998, TREC.