Design And Implementation Of A Natural Language Question-Answering System

With the development of computer and Internet, distance education has been more and more widely accepted in recent years. Comparing with conventional education, distance education breaks the limitation of time and location. No matter when, and no mater where you are, you can share the abundant educational resources on Internet. As a necessary part of distance education platform, question-answering system is very important to improve the quality of teaching. After analyzing shortcomings of existing question-answering systems, we designed an Intelligent Question-Answering System (IQAS) which can answer questions asked in natural language.Actually, the process of answering a question is searching answers relative to the question. Firstly, this paper discusses the key techniques of Information Retrieval, including Chinese segmentation, HowNet and Information Retrieval Model. Secondly, functions the system provides and how to implement these functions are introduced. Thirdly, search process based on natural language, which is the most important part of the system is described in detail.This paper adopts Chinese segmentation algorithm based on both maximum matching (MM) and reverse maximum matching (RMM), and expands query by analyzing relations between words.In our system, questions and answers are stored separately in question database and answer database. Considering the special structure of document set, a vector space search model based on text segment is adopted. An improvement on the traditional TF-IDF formula is made, and local similarity and global similarity are combined to sort search results. Experiment results show that this model has higher precision and recall than traditional Vector Space Model (VSM).At the end of the paper, we summarize features of the system, and point out the shortage which we will improve on in future work. We prospect that question- answering system will be more intelligent and independent.