Research And Design Of Intelligent Question Answering System Based On Data Mining

With the constant deepening of network teaching research, online communication between teachers and students becomes more and more popular in many teaching systems. Whereas with the increasing of information, the traditional way that teachers answering all the questions of students can not meet the needs of students'study. Further more, teachers can't be online all the day and have not so much energy to answer students'questions one by one. But students need their questions to be answered timely and accurately. So, an intelligent auto question answering system becomes the urgent need of teaching system, and it should have a friendly natural language interface for exerting its functions in network teaching.With the constant development of data mining in various fields, people begin to take great effort to explore new application area in recent years. The paper applies the data mining algorithm into the QA system, puts forward a set of scheme about question answering system based on data mining algorithm and realize it. The aim of the scheme is to give up some defects of current question answering system and get an efficiency QA system. The scheme is in detail as following:Firstly, all the information including the questions, answers and remarks from some teaching website is searched and saved.Secondly, the best answer is abstracted from the many answers to every question. The one to one QA pairs will be gotten and saved by classes. By this way a comprehensive and accuracy QA database can finally be formed that can be used to data mining. The improved association rules algorithm is applied to calculate the similarity of documents and text clustering algorithm. By this method, the precision and recall will be more exact and the more accurate answers will be gotten. By using the improved association rules algorithm into the every class after text clustering, the more accurate association table for extracting the better answers from the database can be gotten.Finally, answering the questions. The questions users asked are analyzed and orientated quickly to some class, and then the best answers from the class are abstracted and returned to users. The operating questions and answers are saved into data warehouse using for the second step.Experiments show that the precision and recall will be more improved and the system has merits such as intelligence, continuous self-renewing ability, saving store space and improving executive efficiency etc.