论文部分内容阅读
With the development and penetration of Internet in China, online social network (OSN) is playing an important role in information diffusion.Sina Weibo is becoming more and more popular because of its efficiency,convenience and easy accessibility.Although there are a lot of studies about information diffusion on Sina Weibo,most of them focus on making models of diffusion or popularity prediction.However, it is hard to evaluate how accurate these models are.In this paper, we mainly study the cascades patterns in Sina Weibo.We observe that posts have a weekly periodicity, and the popularity of posts drops with power law The peak period of posting is also different on weekdays and weekends.We also make statistic analysis on the size and patterns of cascades.On one hand, the cascade sharps of Sina Weibo are much more complex than other social network and there is no effective method to distinguish isomorphism;on the other hand, the number of cascades in Sina Weibo is very large;the work of identifying structure and counting the frequency is even more difficult.We propose a method and use indegree and outdegree sequence to count sharps of graphs.The results indicate that most of diffusion networks are star or chain but not the tree.There are also more self-circles and close graphs in Sina Weibo.It manifests that it is more likely for users to retweet their own tweets or retweet reciprocally in Sina Weibo.The low cost and easy accessibility may be the main reason for this phenomenon.These observations can provide deep insight about information propagation.