Community Detection Algorithm Based On Vertex Influence In Semantic Social Network

The semantic web was born in 1998,which is proposed by Tim-Berners.Comparing to the traditional networks,semantic web improves the intelligence of the network,makes up the deficiency of monotone network functions.Therefore it is widely used in intelligent information retrieval,distributed computing,and large-scale knowledge systems,etc.However,in recent years the network has become increasingly complex,and the community detection with semantic web has attracted people's attention.Most of the present community detection algorithms are based on the topology relation in the target network,which find community through the relationship between the nodes and edges.This kind of algorithm can't mine the deeper information contained by nodes,because the research on traditional algorithms has only stay in the connection between the nodes and edges,which means the traditional community detection algorithm has many significant limitations.This paper is based on the community similarity of community discovery thinking,first we use the semantic topic semantic factor nodes in social networks to evaluate the topic information of the node weight by the main components estimation method in statistical.Then we sort the node according the weight in order to find out the most influential nodes in network as the initial community center nodes.At last we calculate the similarity between the remaining nodes and each community center respectively,and divide the remaining nodes into different communities according to the calculation result.This method tries to make full use of the semantic information in the network,and ensure the quality of community detection.In the experimental stage,this paper selects the network of bottlenose dolphins as experimental case,then tests the algorithm results in different initial conditions,and evaluates the results according to the quality function.Experiments show that the algorithm can effectively use the semantic information of the nodes relative to the traditional algorithms,and verify the feasibility of the algorithm for semantic social network community detection.