Research On Community-based Social Network Influence Maximization Algorithm

With the development of Internet technology,researches on social networks have attracted more and more attention from scholars.The analysis of maximizing the influence of social networks is one of the research hotspots.Finding a small number of influential nodes in social networks will eventually affect more other nodes under a certain propagation model.Social network influence analysis plays an important role in advertising marketing,public opinion prediction and monitoring.At present,the influence maximization algorithms are mainly divided into two categories: heuristic algorithm and greedy algorithm.The former is friendly in terms of time complexity,but the quality of the excavated seed nodes is unstable and the influence range is not theoretically guaranteed.Although the latter guarantees the quality of seed nodes,the time complexity is too large,and it is not suitable for the large-scale social network impact node mining nowadays.This paper introduces community structure,an important attribute of social network,into the analysis of maximizing impact.Firstly,presents an improved community discovery algorithm for label propagation,which can solve the disadvantage of unstable community discovery results in traditional label propagation algorithm.Secondly,presents a heuristic and greedy algorithm to maximize the impact of social networks.Firstly,Web ranking algorithm is used to calculate the potential influence of nodes in each community,then some nodes with high potential influence in each community are screened out,and greedy algorithm is used to further screen,so as to mine the influential seed nodes in the community and regard them as global candidate seed nodes.The algorithm also chooses some boundary nodes that connect multiple communities as candidate seed nodes.Finally,the greedy algorithm is executed once again on the global network for all candidate nodes,and the final seed nodes are obtained.Chooses real data sets and makes a comparative analysis of the proposed community discovery algorithm,impact maximization algorithm and several related classical algorithms.The experimental results show the superiority and feasibility of the proposed algorithm.