Optimizing influence spread in multilayer networks: A layer-weighted budget allocation and community-based local dominance approach

The influence maximization problem in multilayer social networks entails selecting seed nodes from each layer under a budget constraint to optimize the overall influence spread. However, the insufficient considerations of topological heterogeneity and cross-layer propagation synergy in existing methods result in imbalanced resource allocation and unsatisfying influence spread easily. To address such challenges, a cross-layer independent cascade model is designed to capture both intra-layer and inter-layer diffusion dynamics. Furthermore, a layer-weighted budget allocation and community-aware local dominance (LWCD) approach is proposed to address the issues of topological heterogeneity and optimal resource allocation in cross-layer propagation. It involves a three-stage process: a layer-weighted assignment method is introduced, where k-core centrality and jaccard overlap are used to quantify the importance of each layer; based on the computed layer weights, the infomap method is applied for community detection, and the budget is allocated proportionally to community sizes; finally, high-potential seed nodes within each community are identified using a local degree metric that captures node influence. Extensive experiments on synthetic and real-world multilayer networks confirm the effectiveness and stability of the proposed LWCD. Compared with state-of-the-art algorithms, LWCD achieves an average improvement of 60.68% in eight networks in terms of influence spread. The complete source code has been made publicly accessible at https://github.com/xiaogoudaidai/LWCDalgorithm to facilitate reproducibility and further research.