TREC 2004 Genomics Track Experiments at IUB

This paper describes the methods we developed for the three tasks of the TREC Genomics Track, i.e., ad hoc retrieval, triage, and annotation tasks. For the ad hoc retrieval task, we used the classic vector space model and studied the use of query expansion and pseudorelevance feedback. Our submitted runs obtained a MAP of 0.183. For the triage task, we adopted a naı̈ve Bayes classifier trained on MeSH terms and used gene names as filters to rule out false positives. The obtained normalized utility score was 0.435. For the annotation task, we focused on document representation and applied a variant of the kNN classifiers. One of our submitted runs produced an F1 score of 0.561, ranking first out of 36 runs submitted for the annotation task.