Research On Answer Extraction For Internet-Based Question Answering System

With the rapid development of Internet, Internet knowledge resources is also in constant expansion. However, webpage so complicated that people are unable to quickly get the information they want. The emergence of search engines, although to some extent alleviated the problem, but the effect is not just as one wishes. The strong demand for rapid access to information, more and more people into the research of automatic question answering system. Automatic question answering system attempts to use without the aid of any human, natural language understanding, information retrieval, information extraction technology, the automatic way to answer user questions.This study Webpage extraction method based on common, proposed one kind based on the "DOM page template and feature information combining extraction" method, in particular to know the candidate question sites such as extracting the user questions and corresponding answers from Baidu set, which is equivalent to the FAQ library from the local extension to Internet, increase the breadth and scope of knowledge, question and answer, the restricted domain question answering system is no longer confined to specific areas of knowledge and then, using Chinese sentence similarity computation thought to find the most similar questions corresponding to the answer returned to the user, so as to realize the automatic answer domain.This article from the Internet question answering system based on background research to start, first elaborated the platform of research, and then a detailed analysis of the calculation of WEB information extraction and Chinese file similarity, finally focus on answer extraction system to do the related implementation.