The effectiveness of dictionary and web-based answer reranking

Chin-Yew Lin
2002 Proceedings of the 19th international conference on Computational linguistics -   unpublished
We describe an in-depth study of using a dictionary (WordNet) and web search engines (Altavista, MSN, and Google) to boost the performance of an automated question answering system, Webclopedia, in answering definition questions. The results indicate applying dictionary and web-based answer reranking together increase the performance of Webclopedia on a set of 102 TREC-10 definition questions by 25% in mean reciprocal rank score and 14% in finding answers in the top 5.
doi:10.3115/1072228.1072254 fatcat:jvcahqbkjngpno2feusl6mvjh4