Design And Implementation Of Automatic Question Answering System

With the improvement of teaching resources and the development of network technology,the center of teaching has changed from teachers' teaching to a new teaching mode based on students' independent inquiry.Under the impetus of students' subjective initiative,they find problems and put forward problems.The enthusiasm for exploring,solving and absorbing knowledge has reached an unprecedented height.How to solve the problems encountered by students after class efficiently has become a difficult problem for more and more teachers,and the automatic question answering system emerges as the times require.Automatic question answering system is an important application in the field of Chinese information retrieval.As a supplement and continuation of the classroom,it can be eliminated when students interact with the system.Students are nervous,shy and afraid,and look for answers to questions in a relaxed,free and positive manner,which plays an important role in students' psychological development and autonomous learning.The automatic question answering system effectively reduces the workload of teachers,frees teachers from the heavy and repetitive answer work,and focuses on improving students' learning efficiency,at the same time,it promotes students' independent thinking and individualized learning.In addition,it can promote the development of online network teaching.Through the analysis of problem records,teachers can find some weak links of students,and provide reference for improving teaching methods and formulating new teaching strategies.This paper compares the research status of automatic question answering system at home and abroad,summarizes the system structure and running mechanism of Lucene,and compares the retrieval performance of Lucene full-text search engine,SQL like clause and MYSQL database.Lucene full-text search engine is selected as the retrieval tool of automatic question answering system.Lucene full-text search engine not only has strong retrieval ability,but also has fatal weakness,because there is no perfect middle segmentation module in Lucene.And the accuracy of Chinese word segmentation in the system itself can not meet the standard of answering questions.In this paper,the accuracy of word segmentation in Chinese is not up to the standard of answering questions.On the basis of referring to the word segmentation dictionary mechanism,the advantages and disadvantages of the three word segmentation mechanisms are compared,and a dictionary mechanism based on double character hash index is proposed.Based on the considerationof the scope of use,the difficulty of operation and so on,the forward maximum matching algorithm is selected as the Chinese word segmentation algorithm of the system,and the algorithm is improved.The improved word segmentation method no longer needs to set the maximum word length artificially.Can automatically get the maximum word length in the process of running.The improved word segmentation algorithm is applied to the architecture of Lucene system,which is combined with the statistical method to make it have a certain effect on cross ambiguity.According to the characteristics of automatic question answering system,this paper puts forward the method of automatic question answering.This paper makes use of natural language processing technology to support the system model of specific subject problem,analyzes and discusses the key technology,and designs and implements the important functional modules.The system understands the natural language entered by the user and returns the answer with the highest similarity.It is proved by experiments that the automatic answer system has good precision and practicability.