Design And Implementation Of A Universal Automatic Question Answering System For Limited Domain

With the popularity and rapid development of Internet,the amount of information on the Internet has increased explosively.How to search the information that we need quickly and accurately in huge amounts of information has become an urgent problem.The information returned by the traditional search engine is too cumbersome to meet the needs of the user.On the basis of traditional search engine,automatic question answering system integrated natural language processing knowledge,and returned answers are more precise and fast,which can better meet users'retrieval needs.Automatic question answering system can be divided into automatic question answering system oriented to restricted domain and automatic question answering system oriented to open domain.The restricted domain question answering system only deals questions in a particular domain and limits the scope of the user's question.The open domain question answering system does not have this limitation,it can handle problems in any field,and users can ask questions at will.However,due to the complexity of natural language,the automatic question answering system for open domain is not very good so far,so the system in this paper is oriented to the restricted domain.On the other hand,the traditional restricted domain automatic question answering system can not be changed after the selection of the domain,and the reusability is poor.This article designs and implements a domain change scheme,and it makes the system of this article can change the domain and have the generality.To improve the accuracy of sentence similarity computation in automatic question answering system,this paper proposed a new TextRank-RD algorithm based on vector space model.The algorithm assigns the node a initial value based on three factors:whether the domain dictionary contains the word,whether the word is a none,the position of the word.And it uses a graph model that assigns weights according to the importance of nodes rather than a graph model that assigns weights equally.The experiment results show that the algorithm improved the accuracy of sentence similarity calculation compared with the TF-IDF algorithm based on vector space model,and is of important significance to improve the efficiency of the automatic question answering system.In this paper,we combine the domain dictionary to improve the effect of the word segmentation,and propose a new method of calculating sentence similarity.Moreover,this paper presents a set of Web information extraction mechanism and FAQ knowledge base update mechanism,enabling the system to achieve better results in the restricted domain,and can be improved with time.At the same time,the method of importing the domain dictionary and the initial FAQ library through the administrator makes the system can change the domain and is versatile.