Analysis Of Information Diffusion Networks Based On Group Chats

The rapid development of online social applications makes people increasingly rely on social tools to spread information.It has attracted many researchers to study the behavior of information dissemination in online social networks.However,previous studies only focused on the networks with explicit dissemination behavior.Few studies paid attention to the networks without clear information dissemination characteristics.In this thesis,we study the problems of group messages propagation within and between groups,try to find the macro and micro laws of information dissemination and the strategies that can promote or inhibit the propagation of information.Aiming at solving the problem of mining message propagation network within chat groups,a heuristic method based on group message context is proposed.First,we analyze the communication characteristics of group members and design corresponding heuristic rules for different group message features.Then,we summarize the context of group messages and use the heuristic rules to infer the possibility of information propagation between two messages.Finally,we traverse the group message and build a messaging propagation network.The experimental results show that the proposed method can reveal the message propagation network and identify the social role of group members through the nodes' network structure properties.Aiming at solving the problem of mining message propagation network between chat groups,a method based on group message content and sender characteristics is proposed and the Monte Carlo method is used to predict the message propagation path.First,we aggregate the messages into a plurality of propagation events.Then,based on the similarity of the message and whether the message sender is a joint group member feature,we infer the possibility of propagation behavior between groups and build the propagation network between groups.Finally,the Monte Carlo method is used to simulate the information propagation process.We analyze the propagation path morphology of different nodes and the influence of the network topology characteristics on the propagation range and speed.The proposed method can accurately mine the message propagation characteristics within and between groups,help people better understand the regularity of information dissemination,identify the influential members of the network,predict the speed and scope of information dissemination.The results of this study may have an impact on many fields,such as information dissemination,business marketing,information recommendation,social management,etc.