Research On Community Detection Method Of Location Based Social Network Based On Dynamic Data Field

With the continuous development of the Internet technology,social networks such as Facebook,Twitter,WeChat,and Weibo have become important social media in everyone's life.At the same time,with the continuous improvement of GPS technology,social networks and location services gradually merged to form a location-based social network,namely LBSN(Location-based Social Network)which fully integrates the online virtual social environment with the offline real-world location environment through the user's check-in behavior.In the study of social networks,community discovery can understand the closeness of connections between users,obtain social relationships and social roles among users,and help to understand the characteristics of social network topology and the evolution of individual relationship behaviors within the community.In the location-based social network,the user's friend relationship is reflected in the topology of the social network,which is affected by the interaction behavior between users,and the geographical location information generated by the user's sign-in behavior is the embodiment of the offline social behavior.By adding geographic location attributes,real geographic location information can be used as an important influencing factor in community discovery methods,making the divided communities more accurate and cohesive.Most existing community discovery methods only consider the topology of user relationships in social networks or community partitioning based on semantics.These methods are not well suited for location-based community networks with multi-mode heterogeneous features.In response to the above problems,this paper proposes a community discovery method for LBSN:Firstly,according to the network characteristics of the location-based social network,the network model is built.In addition to the most basic user-user social relationship in the social network,the social relationship of the user-geographic information is formed according to the user's sign-in behavior in the geospatial space.At the same time,there is a social relationship between geographical location information and geographic location information in the social network.Then,based on the characteristics of users interacting in the dynamic data field,the concept of contextual trajectory for signing context information is proposed.Then,a ternary relationship matrix with three objects of “time”,“location” and “situation” is established for the situation trajectory,and the “location” information and “situation” information in the matrix are used to calculate the location similarity and the topic similarity.The similarity of the situational trajectory can be obtained by combining the position similarity,the topic similarity and the user relationship.Finally,based on the obtained similarity between the user's situational trajectories,the k-means clustering algorithm is used to initialize the community and then perform multiple clustering to complete the community detection.Finally,the feasibility and effectiveness of the location-based social network community detection algorithm proposed by the text is proved by experiments.