Maximizing the Influence in Dynamic Social Networks: An Entropy-Based Linear Threshold Model

Various communication methods allow modern people to communicate with each other more frequently, resulting in a more diverse network structure in today’s social network due to its own dynamics. Compared with the influence maximization under the static network, which has been carried out a lot of related research, the dynamic network’s influence maximization has more practical significance. The dynamic network, which is closer to the evolution of real social network, becomes more challenging for its influence maximization problem due to the variability of its network structure. In this paper, the traditional thresholding model is improved to make it feasible to calculate the threshold at each node and make the model can be used in dynamic networks. By proposing related theorems, we propose a new node seed selection strategy, which can select new nodes that are beneficial to the propagation between nodes by calculating the node influence. Through comparative experiments with five different algorithms, we select two real datasets with large structural differences to generate directed and weighted dynamic networks. The experimental results show that the propagation range caused by our method during the process and at the end of the propagation is much larger than that of other comparison algorithms.