User Knowledge Contribution Willingness Prediction In Social Q&A Community Based On User-Question Matching

In the era of Web 2.0,user generated content has greatly enriched the information resources on the Internet.With the voluntary knowledge contribution of a large number of users,it meets the personalized information search needs of users.Social Q&A platforms provide users with a valuable platform for information sharing and search.Additionally,in social Q&A platforms,users can interact with each other by commenting,voting and etc.,which increases the sense of community belonging.The knowledge in social Q&A platforms is similar to public goods,which leads to a lack of willingness of users to contribute knowledge.Many Q&A communities are faced with the problem of low response rate.However,it is one of the core values of social Q & A platform to provide high quality answers to questions and attract users to participate in discussions.Therefore,it is of great significance to quantify users' willingness to contribute knowledge and make full user of the swarm intelligence of community members.It is not only helpful for users to obtain useful information but also necessary for the sustainable development of community.Many prior works have been done on understanding user knowledge contribution behavior and detecting expert users in social Q&A platforms.In the past few years,research on user's knowledge contribution behavior mostly applies empirical analysis to explore the factors that affect user's knowledge contribution intention from a macro perspective.Most of these expert finding algorithms are able to recognize effectively individuals with the required knowledge to answer a specific question.However,just because people have the capability to answer a question,does not mean that they have the desire to help.In this paper,we quantify a user's willingness to contribute knowledge to a specific question,in order to improve the enthusiasm of knowledge contribution of community members.It provides a method to connect questions with users who are willing to contribute answers,optimizing the question recommendation mechanism of social Q&A platforms.Invitation mechanism is a unique way for question to be answered.We experiment with millions of user behavior data in Zhihu,a representative social Q&A platform in China.Based social cognitive theory and similarity assessment in recommendation system,we use a systematic process to combine the features of three dimensions: user,environment and similarity,consisting of user attribute,question attribute and user's historical behavior.Finally,27 features are extracted from original data set through statistical analysis,social network analysis,text analysis.We build a classification model based on Light GBM?XGBoost and BP Neural Network to predict the probability of users to answer the given question.The classifier demonstrated that we can predict the answer probability of users with these features,and the fusion of multiple models' results can further improve the prediction accuracy.By analyzing some predictors in our model,we notice that user's knowledge contribution intention is affected by the user's characteristics,problem characteristics,and user's internal interest.The user's interest is positively related to knowledge contribution intention,and the relations between part of the features and knowledge contribution intention are U-shaped.This study quantifies users' willingness to contribute knowledge to specific questions,further analyzes the relationship between important features and users' willingness to contribute knowledge.It enriches the research on knowledge contribution behavior,and helps to understand user knowledge behavior.Then,text analysis and social network analysis are used to evaluate the user's interest,by which this paper establishes the foundation for the future research.In practice,this paper puts forward some suggestions for knowledge search users and the operation of platforms.