Research On Question Text Classification And Answer Extraction Technology In Automatic Question Answering System

As an important part of the after-sales service of enterprise products,customer service has a wide range of applications.With the great development of the network,a variety of instant messaging software greatly facilitates people's daily life,WeChat as the most used communication software,it has lots of customs and WeChat public number provide the api for developing again.so WeChat to build the customer service system in the cost of development,promotion and user experience has great advantage.The automatic answer and answer function in WeChat customer service system is a very important module,on the one hand,it reduces the work of customer service,on the other hand,it can make users get faster service.The automatic question answering system can be well applied to the customer service system to improve the service quality and efficiency.Based on the automatic question answering module of WeChat customer service system,this paper studies two important technologies of question answering module,one is the problem text classification technology,the other is the answer extraction technology.The overall content of this article is arranged as follows:This article first introduces the overall framework of the WeChat customer service system and explains how it works and the relevant steps of its operation.The next introducing the problem analysis module,retrieval module and answer extraction module.The above functional modules are implemented in the system.Then the paper analyses the problems text classification method of text analysis module and answer extraction method in detail.In text classification,the traditional machine learning methods existed some questions.For examples,the feature extraction problems of such as extraction of deep syntactic and grammatical features,feature of sparse.Deep learning method can automatically extract features but need a large amount of training data,The shallow linear model has strong memory ability.This paper combines the advantages of them to improve the text classification model.In the designer of the answer extraction module,this paper make it as a classification problem,aiming at the problem of feature extraction of traditional machine learning method,first using convolutional neural network and recurrent neural network to get question and answer ‘s features.And combined with the syntactic features that inputed to an ordinary neural network to build framework and complete answer extraction module.This paper proves the effectiveness of text classification and answer extraction method by comparing with other existing methods,and showing the experimental results.Finally,we implement the customer service system based on WeChat.The problem classification technology and answer extraction technology are applied to the automatic question answering module of the system,making the whole customer service system more efficient and intelligent.