Research And Application Of Influential Nodes Identification Algorithm Base On Gravity Model

The Influence Maximisation problem aims to find the most influential nodes in a complex network as a set of seed nodes,and to treat these seed nodes as the medium that can trigger the propagation of large-scale cascading information,so that the influence can be spread to the widest extent in the network.This problem can be widely used in important application scenarios such as marketing,personalised recommendation and opinion monitoring,and has become one of the key research directions in the field of big data analytics.In this paper,we propose two improved algorithms based on the traditional influential node identification algorithm in order to find the potentially most influential nodes accurately and quickly,and apply the algorithms to public security big data analysis projects,with the following main work:1.Propose an influence node identification algorithm WGEC based on the law of universal gravitation.The traditional gravity-based influence node identification algorithm assigns the same influence range to all nodes when calculating the influence of a node and ignores the information about the node’s neighbors,which leads to biased calculation results.The WGEC algorithm defines a dynamic truncation radius for each node and introduces the potential weights of connected edges,which not only accurately calculates the influence range of each node,but also quantifies the importance of different connected edges The WGEC algorithm not only accurately calculates the influence range of each node,but also can distinguish the different influences generated by the connected edges between nodes by quantifying their importance.The effectiveness and accuracy of the proposed algorithm is verified by combining the classical influence propagation model and evaluation indexes with comparison experiments on several data sets.2.Propose an influential node identification algorithm WGEC-Vote Rank based on the law of gravity and voting mechanisms.The original Vote Rank algorithm ignored the diversity of the nodes themselves,gave all nodes the same voting ability in the initialization phase,and assigned the same amount of decay to all nodes in the update phase,without fully considering the intervention of the actual influence of the nodes on their voting ability.Therefore,the WGEC-Vote Rank algorithm fully integrates the diversity of nodes,combines the node’s own diffusion ability to initialize its voting ability,integrates the voting calculation scores of neighboring nodes within the truncation radius,and uses the effective distance between that node and the seed node to dynamically determine the decay factor of voting ability.The validity and accuracy of the proposed algorithm are verified by conducting comparison experiments on several data sets.3.The influence maximization algorithm proposed in this paper is applied to the actual "public security big data analysis" project to build a network of people involved in the case,dig out the core people in the criminal groups,and reveal the bridge nodes between the criminal groups.