Research On Communtiy Detection Based On User Interaction Behaviors For Online Social Networks

With the development of Internet and mobile devices,more and more people choose to join various online social networks.Different from social activities in reality,online social networks break the limitation of physical distance and time and have strong interaction and timeliness.People in online social networks also form various communities based on their own interests and hobbies.The research of user community structure has important scientific and commercial value.Most existing online social networks rely on the analysis of people's online relationships,such as friends or followers.However,many existing studies have shown that user interaction(such as comments or replies)is a better reflection of user behavior than relationships.To this end,this study designed a corresponding online social network data acquisition system for user interaction,and proposed a community discovery method based on user interaction information.In this study,interaction behaviors in online social networks are analyzed and studied,and the cascading characteristics of relevant interactions are studied in depth,and user interactions are analyzed from the perspective of the cascading characteristics of interactions.In order to discover user communities from online social networks more truly and accurately,this paper designs an online social network community detection algorithm baed on cascading analysis.The cascading relationship between interactions means that interactions can target at other interactions or be targeted by other users.The cascading nature of interaction is ubiquitous in online social networks and an important feature of user interaction.Main research results of this paper:(1)Based on the observation and research of user interaction in online social networks,the cascading characteristics of interaction are analyzed.Through the cascading nature of interaction,an event graph-based representation of the interaction cascading relationship is proposed,which is used to depict the correlation and frequency of interaction between different users.(2)A corresponding data acquisition system of online social network is designed for the interactive behavior of users,which is used to obtain the interactive information between users and the cascading relationship between interactions in the online social network.This method mainly considers the cascading relationship between interactions,the reasonable distribution of users and the integrity of user interaction data.On the basis of data collection,a comparative analysis is made on user interaction data and relationship data.The results show that there is a correlation between user interaction and social relations,but the interaction is not limited to social relations,and is also affected by other interaction behaviors.(3)This paper proposes an online social network community discovery method based on interactive analysis.This method uses user interaction data and extracts chainintensive user groups from the event graph to build the hypergraph.Finding user overlapping communities by looking for user groups with high similarity in the hypergraph.The performance of the proposed community discovery method is evaluated from a number of indicators through testing and verification in three real online social networks.Experimental results show that the proposed method can effectively discover active communities in online social networks.Compared with the current mainstream community discovery methods,it has better detection accuracy and more stable results on multiple data sets.