The Entropy-Based Hop Scheme for Influence Maximization in Dynamic Social Networks

The influence maximization that selects the nodes to maximize the influence spread has been becoming a hot research topic. The classical algorithms based on static network are difficult to solve it in dynamic network where nodes and edges are updated all the time. In this paper, we propose a novel hop approach based on information entropy, which selects the optimal set according to the local topology information while activation probability calculated by improved hop approach in dynamic linear threshold model. This study uses the information entropy to get the unique threshold to reflect the heterogeneity of nodes. In consideration of a node that attempts to activate its neighbors in a static snapshot, the improved hop approach is used to update only the activation probabilities of nodes affected by newly added activation nodes in set of activated nodes. This paper also proves and deduces the relevant theorems and formulas in proposed method. The effectiveness of our method is demonstrated by comparing experiments with five selection algorithms on three real datasets. Through five different sets of experiments, it can be found that our method can get the best results in terms of influence spread and speed of propagation with specified budget.