Research And Design Of The Automatic Question Answering System Based On Full-text Retrieval

The development and maturity of Automatic Question-Answering System (AQAS), that is a subsystem of E-education, depends on the maturity of the involved technologies and the synthetical use of them. Chinese-based fulltext retrieval (FTR) is a key technology in AQAS. This thesis conducted a literature review on national-international latest improvement of FTR and AQAS. Based on the emerging demands in web education as well as the state-of-art development of relevant technologies, research and design web AQAS has been done. Based on analysis of some existing text retrieval models and the evaluation of FTR system performance, the thesis introduces the structure and central data stream of Dotlucene, which is FTR engine, and a method based on DotLucene has been selected as Chinese-based FTR engine in develop local AQAS. Then, this thesis brings forward a new method of natural language processing in limited field. The method detoured difficulties of natural language participle, by matching the natural language with keyword list and by substituting the keyword combination, realized to answer the question that is asked with natural language. Lastly, in this thesis, ASP.Net and ADO.Net have been discussed and demonstrated in developing framework of AQAS and realized AQAS based on FTR. As a result, this thesis's method and technologies are found effective and can be applied to AQAS of E-lessons.