Read and reply behaviors in a BBS social network

Understanding user´ read and reply behaviors are important both for the study of online social network properties and the modeling of information dissemination and trend prediction. Our dataset is crawled from a famous Chinese BBS social network. By analyzing the read and reply datas of users, we observe that distributions of discussion size and user participation level follow the power-law distribution, agreeing with former empirical and analytical investigations. But the distribution of user interest does not exhibit a power-law property. Statistics from hot posts show that read and reply behaviors vary among different topic categories. For every 100 users interested in a post, one or two of them in average would join the discussion. In addition, investigation concerning time property of replies shows that decay functions with periodical fluctuations are suitable to describe the short-term evolution.