Question Answering by Passage Selection (MultiText Experiments for TREC-9).
Charles L. A. Clarke, Gordon V. Cormack, D. I. E. Kisman, Thomas R. Lynam · 2000
Introduction MultiText has participated in TREC each year since TREC-4 [3, 2, 7, 8, 6]. For TREC-9 we concentrated our efforts on the question aaswering (QA) track and also submitted runs for the Web track. The MultiText system incorporates a unique technique for arbitrary passage retrieval, which was used in all of our TREC-9 experiments. The technique efficiently locates high-scoring passages, where the score of a passage is based on its length and the weights of the terms occurring within it. Passage boundaries are determined by the query, and can start aad end at any term position. If a document ranking is required, the score of a document is computed by combining the scores of the passages it conta/ns. Versions of the technique have been described in our previous TREC papers and elsewhere [4]. Our TREC-8 paper [6] provides a concise overview of the current version. Our TREC-9 QA experiments extended our TREC-8 work. For TREC-8 we simply stripped stopwords from the question, app