The Analysis of Influencing Factors of Information Dissemination on Cascade Size Distribution in Social Networks
Jian Dong,Bin Chen,Liang Liu,Chuan Ai,Fang Zhang
DOI: https://doi.org/10.1109/access.2018.2871145
IF: 3.9
2018-01-01
IEEE Access
Abstract:The cognitive behavior of online user groups on information promotes the dissemination of information. The trajectories of information dissemination in social networks can be described by treelike cascades, and the distribution of the sizes of these cascades can capture the distribution of popularity of a social network. Numerous studies have shown that the cascade size distribution follows fat-tail distributions, including power-law distribution and bimodal distribution; however, the underlying characteristic of this highly skewed distribution lacks quantitative experimental analysis. Based on the stochastic epidemic-like information dissemination model, namely, the susceptible view forward removed model, this paper explores the impact of the content attractiveness and influence, and information source on the cascade size distribution through computational experiments. On the one hand, we find that when the mean value of the information influence and attractiveness is small, the cascade sizes follow a power-law distribution, and the larger the variance, the heavier the tail. On the other hand, the more random the distribution of information sources in social networks, the smaller the slope of the power-law cascade size distribution. Our findings quantitatively reveal the causality of power-law cascade size distribution from computational experiments, clarify the role of information attractiveness, influence, and sources on the distribution of popularity in social networks.