The Research Of Answer Extraction Algorithm In Chinese Question And Answering System

"We are drawn with information, but we are eager for knowledge." The word by John Naisbitt, is giving a vivid picture of the awkwardness and bemusement of modernists facing the explosion of information. On one hand, the expansion of Internet brings the even an individual corporation the enormous information. On the other hand, the situation that we human being are lack of efficient toolkit to acquire useful information. Although the companies of web search like Google, Baidu, Yahoo, and so on, and the enterprise search corporations like IBM, Endeca, are trying their best to improve the performance of their own products. However, honestly speaking, the development of original search engine has failed to meet the requirement of people so far. So, Question and Answering System is created to solve this problem.Based on the original search engine, Question and Answering System adds NLP (Natural Language Processing) into the architecture. The aim of QA is to raise the search engine from the syntactic layer, to the semantic and pragmatic layer of understanding the aim of users. In this paper, we introduce the all the modules of Chinese Question and Answering System, the main search approaches and algorithms. Other than that, the focus of this paper is on the answer extraction module of Chinese Question and Answering System. In this module, the paper raises two methods, one is three layers based on Wikipedia, with keywords and syntactic pattern and the other is constructing semantic community network based on Wikipedia, to score, order and reduce redundance of answers. In this paper, the two methods is compared by the results of experiments, and draws the conclusion that the algorithm based on semantic network is the trend of complex question answer extraction research. At last, the paper gives a quick look at the Answer Path participating NTCIR-7, and suggests the work of improvement and research in the future.