Research On Knowledge Map Construction Of Virtual Q & A Community

In the era of Web2.0,with the rapid development of virtual Q & A community,questions and answers community such as Yahoo! Answers,Baidu Know,Zhihu,etc.which have accumulated large archives of questions and answers.These archives deliver large valuable knowledge to users.Browsing is one of the important methods to acquire knowledge.In questions and answers community,question-answer pairs,which includes not only knowledge contents but also knowledge needs,is a new form of knowledge presentation.In order to facilitate browsing the archives of questions and answers and reduce the information overload,this dissertation proposes a knowledge map construction method of virtual question and answer community.The knowledge map reveals the relationship between knowledge needs and knowledge content.For the research of virtual question and answer community knowledge map construction,TF-IDF is used to model the questions and answers firstly.Then the structure of the knowledge map is completed by using the proposed knowledge map construction method: GH2 DSOM.Questions and answers are two dimensions of the knowledge map,the intersection of the two dimensions of the question and answer is the corresponding question-answer pairs.After the construction of knowledge map structure is completed,Label SOM algorithm is used to extract the keywords of each knowledge node of the two dimensions,and the representative question-answer pairs of intersection of the two dimensions of the question and answer are extracted with the representative question-answer pairs extraction method proposed in this dissertation,so that users can browse the knowledge map.In order to verify the feasibility of the knowledge map construction method proposed in this dissertation,questions and answers is obtained from ZHIHU which is knowledge sharing platform.The experimental results show that the performance of the knowledge map structure evaluation is satisfied,the average coverage rate of the representative question-andanswer pairs exceeds the proportion of the extracted question and answer pairs,the effect of consistency is better,and the average user satisfaction is satisfactory.These experimental results verify the effectiveness of the proposed method.Users can find the corresponding and high-quality knowledge content by browsing the knowledge map,and the user can grasp the knowledge of the category by reading the representative questionanswer pairs in a short time.