Question Answering Using a Large Text Database: A Machine Learning Approach
Hwee Tou Ng, Jennifer Lai Pheng Kwan, Yiyuan Xia · 2001
In this paper, we present a machine learning approach to question answering. The task is answering factual questions, where the answers are to be found in documents in a large text database. We trained our system on 398 questions from the Remedia corpus, as well as 38 TREC-8 development questions. We then evaluated our system on 198 questions of the TREC-8 question answering task. Although our learning approach only uses 4 features, we are able to achieve quite competitive accuracy. The results indicate that such a machine learning approach is a promising way to build a state-of-the-art question answering system.