Empirical Analysis on the Users' Reply Behaviors of Online Forums

摘要

Quantitative understanding of human behaviors is of huge significance to uncover the origins of many social and economic phenomena. This paper focuses on users' reply behaviors of online forums. The statistical results show that users' one-day activities and one-post interests both follow heavy-tailed distribution not only on population level but also on individual level. Specifically, we found that users' one-day activities obey a power law distribution with exponential cutoff and users' one-post interests obey a power law distribution. Further, we observe a positive correlation between users' total activities and the power law exponents of one-day activities. In addition, we study the content scatter degree of users' one-day replies and the time scatter degree of users' one-post replies, and find some suspicious user behaviors, which provide new clues to the detection of cyber-space hype and network water army.