Community-Based Memetic Algorithm for Influence Maximization in Large-Scale Networks

Effective information diffusion across large-scale network is key for influence maximization. Recent research has shown a significant surge in interest in modeling, performance estimation, and seed identification across various networked systems. Moreover, a simulation of useful interactions among many significant groups within networks was developed to simulate real-world marketing and spreading information more accurately. A good diffusion model identifies the minimum number of effective seeds capable of achieving maximum diffusion effects across the network. Limited focus has been placed on measuring the strength of seeds in competitive spreading situations. There is a research gap in determining effective strategy for this purpose. This study proposes a memetic algorithm based on a community for large-scale social networks. The proposed algorithm optimizes the influence spread by identifying the most influential nodes among the communities, depending on their inter- or intra-community propagation dynamics. This algorithm combines the concept of genetic algorithm with a reachability-based local search method to accelerate the convergence process. This approach offers a robust method for maximizing the influence of network structure and interactions. An experimental evaluation on real-world social network datasets shows the performance superiority of this community-based memetic algorithm (CBMA-IM) over existing algorithms.