Research Of Specific Domain Question Answering System Based On Internet Information

The explosive growth of all kinds of information on the Internet has caused people to spend a lot of energy when getting accurate information.In order to solve this problem,search engines have appeared,search engines have helped solve this problem,but it also has many limitations.The most important point is that it can only return a series of pages according to the degree of association,instead of a sentence.People still need to spend a lot of time looking for the answers they need.Therefore,people still need a way to obtain information more simply and quickly.At this time,the question answering systemappears and becomes a hot research topic.Among them,the specific domain question answering system has attracted much attention because of its high feasibility and practicality.And the vast amount of network information provides a huge source of information for the question answering system.Therefore,this paper is devoted to the research of the specific domain question answering system based on network information.It is mainly divided into the following three parts.Automatic construction of knowledge base based on network information.The construction of past knowledge bases often requires a lot of labor,and it takes a lot of effort.In order to solve this problem,this paper designs a method of automatic construction of knowledge base based on network information,which is based on the domain word set to collect questions and answers in the encyclopedia and online answer community.The domain word set is obtained by crawling the domain website corpus and extracting the domain words.In the process of domain word extraction,this paper proposes an improved domain word extraction method based on TextRank and Word2 Vec,and achieved good results.Construction of a retrieval question answering system based on network information.Due to the limitations of the content of the traditional question answering system knowledge base,many questions could not find a matching answer.In order to solve this problem,this paper designs a framework of retrieval question answering system based on network information,which will put questions that cannot be answered by the question answering system into the encyclopedia,online Q&A community and search engine to match or search the answers.The domain knowledge base will be expanded when the framework runs.With this framework,the problem of“no response” is basically solved.This article also combines the automatic database construction,designed and implemented a system that can help users build a question answering system quickly and automatically.Construction of a generative question answering system based on networkinformation and deep learning.The traditional rule-based template and retrieval question answering system is based on the existing knowledge,extracting an existing corresponding answer,and does not really understand the question to produce an answer.The real generation question answering system,which was previously used in the open chat field.In this paper,a qualified domain generated question answering model is established based on the Encoder-Decoder framework and BiLSTM.And the Denoising AutoEncoder and the word vector combined with part-of-speech and domain word information are used to improve the performance of the model.The decode combined with improved Beam Search algorithm will be used to improve the output quality.Under the indicator BLEU,all of them improved the effect of the model.