An Analysis on User Behaviors in Online Question and Answering Communities

Community-based Question and Answering, or CQA, has been extensively studied nowadays. Studies focus on the characteristics on users help us understand the usage of the platforms, and pave the way for downstream tasks such as churn prediction, expert recommendation and spam detection. Different from previous works, we focus on analyzing the behavioral differences between different types of users, take the feedback of the community into consideration, and predict users’ future behaviors by their historical behaviors and feedback of the community. In this paper, we conduct a case study on Zhihu.com. We collect data of questions, answers, comments and users from Zhihu.com, make a brief analysis of the usage, divide the users into 3 groups named Takers, Devotees and Participants and study the differences between each type by KS tests, and conduct linear regressions to predict users’ future behaviors using their historical behaviors and feedback from the community. We find significant differences between each type of users, and find several positive correlations between users’ historical behaviors, feedback from the community and users’ future behaviors. By the analysis of users’ behaviors, our research provides a theoretical basis for the follow-up studies.