Research On The Influence Maximization Algorithm Based On Structural Hole Theory

The development of Internet technology promoted the vigorous development of a variety of social networks(Facebook,Twitter,Sina microblogging,etc.),The excavation of rich information data in the social network and the marketing activities through the social network have brought unprecedented challenges and opportunities to the social network research.Internet marketing is marketing model based on personal through social relations on the surrounding friends,family or colleagues to influence the proliferation,which resulting "word of mouth".In the process of influence diffusion,mining the most influential user in social networks becomes critical.In this context,the impact of the issue of maximization of research has become hot.The problem of maximizing influence is to select some nodes from the social network as candidate nodes and make full use of these candidate nodes so that the ultimate impact on the social network is greatest.At present,many algorithms have been proposed to maximize the impact of social networks,such as various greedy algorithms and heuristic algorithms.The aim is to reduce the time cost of maximizing the algorithm and improve the accuracy of the algorithm.However,there are few structural features in the existing algorithms,such as structural node nodes with distinct structural features,and stronger ability to communicate with ordinary nodes.Ignoring the structural characteristics of nodes The effect of maximizing the algorithm is not ideal,and the use of structural features can effectively reduce the time cost of the algorithm.In view of the above challenges to maximize the challenges and problems,this paper started according to the following two aspects.The first is to study the influence maximization algorithm based on structural hole theory.Proposed SG(Structure-based Greedy)algorithm to reduce the impact of the maximum time to maximize the algorithm and improve the impact of the spread of the scope.The basic idea of the SG algorithm is to establish a Laplace matrix for the primitive social network and to determine the structural hole nodes by solving the Federer vector.In this paper,we consider the structural features and influence of the nodes,and filter out the nodes with non-structural nodes and the small influencing factors in order to reduce the candidate selection space.The NewGreedy algorithm is proposed to select the seed nodes which can get the maximum propagation range in the narrow candidate set.In this paper,the experimental algorithm based on structural hole theory proposed in this paper,compared with the existing algorithms,has obvious advantages in terms of algorithm time cost and algorithm result quality.The second is to study the relationship between efficient algorithmic discovery and community division.The main reason that the existing structure hole discovery algorithm is to calculate the structural hole value of the node.In the large-scale social network,the computational cost of the structural hole value is considerable.This paper proposes to reduce the computational cost of structural hole value by means of the idea of community division.Firstly,the original social network is roughly divided according to the similarity of the nodes to get the simple community structure,and then determine the structure hole node according to the community structure,and finally calculate the structural hole node and the similarity value of the existing community as the node structure of the hole value.In addition,based on the premise of the known social network structure,this paper proposes a new community division method based on "two-step" information flow theory SCD(Structure_based Community Detection).The basic idea of the SCD algorithm is to first discover the original social network according to the similarity of the nodes,and then determine the structure of the network node based on the discovered community.Finally,from the structural hole node,according to the"two-step" information flow theory to detect potential societies,according to the potential community and the original simple community similarity found after the new community.In this paper,the correctness and accuracy of the proposed community-based community detection algorithm are greatly improved by real experiments.