A Novel Measurement of the Activation Probabilities in Information Diffusion Model

Abstract—Microblog is one of the most efficient and effective way to spread information. This paper studies the information diffusion in Sina microblog and proposes a novel measurement of the activation probabilities based on Independent Cascade model. The activation probabilities measure considers tweets, relationships, and users, so it is called TRU (Tweets, Relationships, and Users) measurement. We randomly select more than 3 million tweets to analyze and find that multimedia plays a very important role in information diffusion. The relationships contain not only the similarity of interests, but also behavioral momentum between users. The users mainly consider a user’s influence in the microblog network and we apply a PageRank-like algorithm to measure it. The TRU measurement has dynamic and heterogeneous nature. It is evaluated in a network with millions of users collected from Sina microblog. And zombie fans whose followers or tweets are less than five have been cleared away from the network. In the experiment, we do not take the parameters in the formula of activation probabilities for granted, but try different values time after time using different tweets to find out the best values. The result shows that TRU measurement can pretty much describe the actual information diffusion in microblog. Keywords-microblog; information diffusion; activation probability I