Research On Information Diffusion Mechanisms In Online Social Networks

Online social networks are now recognized as an important platform for the spread of information.Studying the information diffusion mechanisms in the online social networks will greatly help us to understand the properties of network structure and the behaviors of the user group.Besides,it is of great significance to national security and social stability.In this thesis,we explore some key issues of information diffusion mechanisms,including diffusion modeling,popularity analysis and limiting the spread of misinformation.We carry out our work on retweeting behavior under multiple exposures,popularity evolution and opinion guidance,taking weibo as an example.The main work and contributions are listed as follows.(1)To understand the retweeting behavior under multiple exposures,we perform an empirical study of the problem on Sina Weibo.We begin with the analysis of retweeting behavior from the perspectives of social influence and homophily.Next we analyze the effect of the order of exposure on retweeting.Finally,we propose an individual interaction model to infer the predecessors.Experimental results show that our model is more accurate than other models.(2)Few efforts have been made to understand the relation between topological aspects and temporal properties of information diffusion.We attempt to describe and explain the process that popularity evolution of hashtags.A temporal analysis highlights the potential effect of topology of diffusion process on the peak popularity of hashtags.We further propose a model to predict the peak time and peak volume.The experimental results show that our model outperforms two related models.(3)We propose an opinion guidance model to limit the spread of misinformation.Based on user's opinion and sentiment,the model recommends messages or other users that have similar viewpoint and positive feeling to current user.(4)A platform for information diffusion analysis is designed.Based on the existing microblogging data crawling system,we implement some methods proposed in this thesis for retweeting analysis,trend analysis and opinion guidance.In the future work,computational efficiency and accuracy of proposed models can be further improved.In addition,dynamical characteristics will be further mined to promote deeper analysis of information diffusion mechanisms.