Topic-Aware Information Coverage Maximization in Social Networks

Zhihang Li,Hongwei Du,Xiang Li

DOI: https://doi.org/10.1109/tcss.2023.3243936

2024-01-01

IEEE Transactions on Computational Social Systems

Abstract:Influence maximization (IM) aims to identify a set of nodes $S$ to maximize the expected number of nodes influenced during the information propagation starting from $S$ . Some works had extended this problem to be topic-aware , where each node is associated with a topic distribution and tends to be activated with different probabilities by different topics. However, whether it is topic-aware or not, IM problem only focuses on the active nodes and overlooks all the inactive ones. Actually, an inactive node may receive the information from their active in-neighbors and become informed. Therefore, this type of nodes should also be considered when measuring the coverage of information propagation. Inspired by this, we formulate a new problem called topic-aware information coverage maximization (TAICM), which aims to maximize the sum of the expected number of both active and informed nodes in topic-aware social networks. Then we devise a heuristic method to solve it. Experiments on three real-world datasets demonstrate that our method can achieve similar or higher information coverage in much less or at least acceptable time than some commonly used IM algorithms.