作者: Xiaohui Bei , Ning Chen , Liyu Dou , Xiangru Huang , Ruixin Qiang
关键词:
摘要: In this paper, we introduce a trial-and-error model to study information diffusion in social network. Specifically, every discrete period, all individuals the network concurrently try new technology or product with certain respective probabilities. If it turns out that an individual observes better utility, he will then adopt trial; otherwise, continues choose his prior selection. We first demonstrate trial and error behavior of characterizes global community structures network, from which are able detect macro-communities through observation micro-behavior individuals. run simulations on classic benchmark testing graphs, quite surprisingly, results show dynamics even outperforms Louvain method (a popular modularity maximization approach) if have dense connections within communities. This gives solid justification model. influence problem dynamics. give heuristic algorithm based detection provide experiments both large scale collaboration networks. Simulation our significantly several well-studied heuristics including degree centrality distance almost scenarios. Our reveal relation between budget advertiser invests marketing strategies, indicate mixing parameter, evaluating structures, plays critical role for diffusion.