Research On Subgroup Discovery Method Based On User Behavior Analysis

Social media is a new type of online media.The main purpose is to encourage users to communicate with each other and increase communication between people.With the rise of many social networks,more and more people are connected by these social networks in the form of a network.Through the use of these new media,they communicate with each other,obtain information,share opinions in real or anonymous name and evaluate event.These operations will generate behavior records and user interaction networks.User behavior analysis emerge,mainly to analyze the interaction behavior between users and their own characteristic attributes,in order to find certain rules to serve the actual application.Subgroup discovery is a descriptive data mining technology,which is designed to identify the description of subgroups in the data set,and to identify specific groups through certain specific user attributes or group structures formed by interaction between user groups.The existing subgroup discovery method has many methods based on graph structure,but does not consider the hidden interaction relationship between nodes,and ignores the influence of the attribute importance of the node itself.Considering the above problems,the impact of node implicit relationship is reconsidered,and the user interaction behavior is quantified and reflected in the graph structure.A Subgroup Discovery based on Behavior Interaction method(SDBI)is proposed.Firstly,the data is analyzed by the method of “prominent data classification”,and some attribute features are extracted for analysis to measure the importance of the nodes,and the remaining important data are sequentially extracted according to the importance of the data.A new node weighting method is proposed.According to this node weighting method,the network formed by the selected data nodes is reflected on the directed weighted complete graph.After that,the SDBI method is used to perform subgroup discovery.The results show that the SDBI algorithm can effectively consider the data non-uniformity in the subgroup discovery process and improve the partitioning accuracy.