Research On The Key Technologies Of Social Network Structure Partition

With the development of information technology,a variety of social networks have had a profound impact on people's life.In the real world,social network data is huge and various,which contains a lot of information.How to mine valuable information in social network has become a hot topic.As an important method of mining and understanding the basic structure of social network,network structure partition has always enjoyed great attention.At present,a lot of related research has been done on the network structure partition,but there are still some problems: 1.The role of key nodes in network hierarchical structure is not clear,and the relationship between hierarchical structure and community structure is ignored.2.The nodes and connections in social networks are heterogeneous,different types of nodes have different connection preferences.How to make a more accurate division of network structures based on hierarchical division is also an important issue that people urgently need to solve.3.There are usually lots of label information besides topological information in social network,how to make more reasonable structure division based on the existing label information of nodes also needs further analysis.In response to the above problems,this paper studies the method of social network structure partition.The main tasks are as follows:1.A method for analyzing the hierarchical structure of social networks based on the deletion of key nodes is proposed.Firstly,a fast node-intermediate computing method is designed to mine the key nodes in the network.Secondly,the iterative deletion of key nodes is used to decompose the network and the decomposition process is performed.The key nodes are the high-level nodes of the network.The disassembled connected branches are used as the community structure in the network.Then the high-level nodes and communities are reordered in the adjacency matrix to visualize the hierarchical structure of the network,and the hierarchical structure model of the social network is obtained.Finally,through experimental verification,the key node deletion method can make a realistic hierarchical structure division for most social networks.2.A diversified network structure partition method based on the minimum description length principle is proposed.Firstly,a network model with both hierarchical and diversity structural features is established.Secondly,the network is described using the model to obtain a smaller network description length based on the principle of minimum description length.To minimum the description length,first divide the network hierarchically,and then decompose and merge the nodes in the low-level network in the direction of the smallest description length.The model with the smallest description length will be used as the optimal partition of the network.Finally,through experimental verification,the diversified network structure analysis method based on the principle of minimum description length can make a more precise hierarchical and diversified structural partition of social network structure.3.An attributed network structure analysis method based on attribute information entropy is proposed.Firstly,propose an attributed network information entropy calculation method based on tag information entropy and structural information entropy of attributed network.Secondly,partition the network initially according to the network topology,then labeling the subgraphs and calculating the information entropy of the attributed network.Merging the nodes with different attributes into the connected subgraphs with the same attributes if the description length reduces.Get the final structure of the attributed network until the description length no longer changes.Finally,it is verified by experiments that the attributed network structure analysis method based on attribute information entropy can make use of the attribute information of nodes to further optimize the structure partition of the network,and at the same time make a corresponding annotation on the network structure.