论文信息 - Answer Selection and Confidence Estimation

Answer Selection and Confidence Estimation

We describe BBN’s Question Answering work at TREC 2002. We focus on two issues: answer selection and confidence estimation. We found that some simple constraints on the candidate answers can improve a pure IRbased technique for answer selection. We also found that a few simple features derived from the question-answer pairs can be used for effective confidence estimation. Our results also confirmed earlie rf indings that the World-Wide Web is av ery useful r esource for answering TREC-style factoid questions.

Scott Miller | Jinxi Xu | Ralph M. Weischedel | Jonathan May | Ana Licuanan

[1] Justin Zobel,et al. Effective ranking with arbitrary passages , 2001, J. Assoc. Inf. Sci. Technol..

[2] Ellen M. Voorhees,et al. Overview of the TREC 2002 Question Answering Track , 2003, TREC.

[3] Richard M. Schwartz,et al. An Algorithm that Learns What's in a Name , 1999, Machine Learning.

[4] Richard M. Schwartz,et al. A hidden Markov model information retrieval system , 1999, SIGIR '99.

[5] Adwait Ratnaparkhi,et al. IBM's Statistical Question Answering System , 2000, TREC.

[6] Jimmy J. Lin,et al. Web question answering: is more always better? , 2002, SIGIR '02.

[7] Scott Miller,et al. A Novel Use of Statistical Parsing to Extract Information from Text , 2000, ANLP.