Research On Expert Finding In Online Knowledge Communities

With the rapid development of the Internet,it follows then that online knowledge communities spring up.Online knowledge community is a kind of network virtual community,specializing in knowledge sharing and seeking.However,due to the differences in user expertise and the influence of spam information,the number of unanswered questions in the community is increasing,which directly affects the development of the community and the user experience.Therefore,it is of great practical significance to find expert users in online knowledge community for the development of knowledge community and the expansion of user services.In this paper,we introduce the research background and related work of expert finding in online knowledge community.At present,most of the existing expert finding techniques are mainly based on the methods of topic model or link analysis.The method based on topic model generally extracts topic or interest distribution from user generated content.The method based on link analysis mainly adopts the link analysis algorithm to analyze the Q&A relationship of users in online knowledge community,and then calculates the social influence of users.Based on the characteristics of online communities such as user generated content and Q&A relationship,this paper studies the method of expert finding in online knowledge communities and proposes a question domain expert finding method.Aiming at calculating the professional level of users,this paper first extracts the Q&A relationship of users in online knowledge community,and based on the analysis of user group effect and user activity,the Leader Rank algorithm is improved to calculate the social authority of users.And meanwhile,this paper adopts the feedback of users to evaluate the quality of users' historical answers.Combined with the quality of users' historical answers and the authority of users,a new expert ranking method named LREF is proposed to calculate the expert scores of users.Aiming at recommending experts to new questions,this paper proposes a question matching model.First,we generate user profiles based on the previous answers,comments of the user and related question information,and extracts the user knowledge tags from user profile.And then,we calculate topic relevance between the user knowledge tags and the new question,and expert users in question related fields will be recommended for the new question by combining with expert scores.Finally,a question domain expert finding method that combines link analysis,user activity and user generated content analysis is proposed.In this paper,we preprocess the original Q&A dataset from Stack Exchange community to form a standardized Q&A dataset.Then,using the standardized dataset,the proposed method of question domain expert finding was compared and analyzed experimentally.Experimental results show that the question domain expert finding method proposed in this paper has better performance and accuracy than the classic expert finding methods under the same background with an amount of data samples.