Maximizing the Impact of Competitive Relation Based on Node diffusion Model

Abstract

The problem of maximizing influence is to select a small number of nodes (seed nodes) in social networks so that they could maximize the spread of influence. In our paper, we have two contributions. First, we consider the impact of competition among products on nodes based on the Susceptible-Infected (SI) information diffusion model and improve it into a Rival Susceptible-Infected (RSI) information diffusion model. Second, the probability of influence between nodes is different based on different products under the same theme with competitive relations, and in the process of the node spreading one of the products, other products under the same theme have a competitive blocking effect on the node. So a new heuristic algorithm— Influence Reconstitution Algorithm (IRA)—is proposed to consider the impact of product competition on the nodes and the distance between the initial nodes. Our algorithm introduces the k-order core competition influence and coincidence rate P. Through the coincidence rate P, the influence of the initial node is reasonably controlled, and the most influential nodes are found in order. Our experiments based on real data sets shows that the IRA algorithm has better competitive effect than the existing heuristic algorithm in the competitive relational social network.