Research On The Important Nodes Mining In Social Networks

Social network is an application of complex networks about the real world,it's a collection of individuals and different social relations,it reflects the mutual activity between different individuals.Social influence analysis is an important research direction of social network,the individual's influence and importance measuring is a hot issue and it has an important role in the field of the spread of the disease and merchandise marketing.Many algorithms have already been proposed,this paper mainly studies the important node mining problem in social network.The existing node influence and importance measuring algorithms are mostly concerned with the node's own attributes,the importance of a node is not only related to its own local attributes,but also related to the global characteristics of the network where the node is located.This paper proposes a KSLD algorithm based on two scales which are the node's local attributes and global attributes to dig important nodes.First the KShell decomposition is used to divide the network layer and the entropy is used to measure the ability of nodes to influence the network layer,then the efficiency and effectiveness of traditional algorithms is balanced and improved based on their advantages and disadvantages,the local degree centrality algorithm is proposed to calculate node's local information,the experiment verifies the validity of the local degree centrality algorithm and the KSLD based on dual-scale,and proves that the KSLD algorithm has better applicability for different types of network than the traditional algorithm.Most existing works on node influence analysis focus on static network,the node influence characteristics based on static network can not be very suitable for the actual network situation because the actual network is evolving.This paper analyses node influence based on the evolution network,the characteristics of information transmission in evolution network is analyzed,and the event attributes of nodes leave community or join new community is found,then the events is defined and a method of evaluating the influence of nodes is given based on this definition,the experiment proposes two indicators which are social index and influence index to sort the nodes,the results demonstrate the feasibility of the method.