An Answer Extraction Method of Simple Question Based on Web Knowledge Library

The answer extraction is an important step in Q&A System and other fields of the Natural Language Processing; the result of answer extraction is directly associated with the quality of Q&A System. In this paper, the method, answer extraction of simple question based on web knowledge library, is realized through computing the similarity of simple question; the question with the most similarity is used as the recommended answer, and questions that the similarity is above 0.5 are used as candidate answers. Experiment shows that the method is feasible, and the accuracy rate is 79.8%. The result of experiment shows that the system is significant for simple informative questions, and the MRR value of tested FAQ is 0.50.