Evaluating The Quality Of Content In Social Q&A Systems

With the development of web 2.0,the Internet has become a more open information space,and the user is becoming more involved in the interaction process of information.It has become a commonly used way to seek answers to solve their confusions from the social Q&A platform based on the user knowledge contribution,which is an important channel for knowledge producing and spreading.But,at the same time,this kind of "public participation" approach presents a greater challenge in the content quality.So,content quality assessment is very necessaryThis paper put aside the traditional questionnaire survey method and put forward a new social Q&A community content quality assessment scheme.It attempts to combine term frequency analysis,topic analysis with sentiment analysis to do some research,which uses the method of data mining and machine learning.After that,by comparatively analyzing the short text about the new major "Library and Information Science" grabbed from the two typical domestic social Q&A platforms-Zhihu,Baidu Zhidao,it verified the feasibility of the scheme.At the same time,the experiment result can also make it more clear about the two platforms' focus on MPLS.Besides,it can also provide a choice for users to better understand the major.The main experimental analysis process is:First of all,analyze term frequency,namely to scrape question and answer data from two platforms respectively,analyze some relevant characteristics,such as high-frequency words/common words.In detail,let a represents the number of common words and b represents the number of common words related to the search word,and we use b/a as one evaluation criteria.Then,topic research,namely to detect topic numbers primarily(For convenience,let m represents questions' topic number and n represents answers' topic number)using hierarchical clustering and do some analysis,using matching degree,namely |m?n|/|m| as one evaluation criteria.As to sentiment analysis,we use the proportion of the text that contains emotional polarity in the answer texts,which is the sum of the proportion of positive and negative texts,as one evaluation criteria to evaluate platforms' answer content.