Multi-Targets Influence Maximization Algorithm Based on Multi-Cascade Model

摘要

The problem of maximizing personalization influence is to use the online social network as the background to target the specific network users and mine the initial influence communication user set that maximizes the impact of the network users. But existing algorithms still rely on traditional traditional models and cannot imitate real information propagation. On the other hand, the algorithm research on multi-cascade model needs to be expanded. In order to better conform to the information propagation in real life and improve the performance of the algorithm, this paper adopts a multi-cascade model to set the user state to an integer value reflecting the influence quantity, instead of the active or inactive two states in the traditional independent cascade model. Users with the same hobbies tend to get together, so you can use this feature to classify the entire target set using clustering. Based on the above, this paper proposes a multi-targets influence maximization algorithm based on multi-cascade model, clustering candidate users, using clustering center as the seed node to spread information, to maximize the impact on specific users. The intensity of the impact on the target user is measured by the frequency of the individual. The comparison experiments on real social networks show that the personalized impact maximization algorithm based on multi-cascade model has better time performance and propagation effect than the algorithm based on traditional independent cascade model.