Dynamic Social Network Optimization via a Hybrid Genetic Algorithm with MLE-Enhanced Fitness

Dynamic social networks present significant challenges for rapid identification of high-impact nodes, which is crucial for effective information dissemination in scenarios such as emergency management and enterprise marketing. This study introduces a hybrid optimization framework, MLE-GA, that integrates maximum likelihood estimation (MLE) with traditional genetic algorithms (GA) to address these challenges. In the proposed two-layer architecture, the outer GA layer performs a global search to optimize candidate seed node combinations, while the inner MLE layer dynamically estimates network parameters in real time, thereby constructing a probabilistic model that guides the evolution of the solution. Simulation experiments, including tests on synthetic datasets and real-world networks like Zacharys Karate Club, demonstrate that MLE-GA achieves an error rate below 5% for identifying high-influence nodes, significantly outperforming conventional MLE approaches. The results confirm that the hybrid method not only effectively distinguishes between high- and low-influence nodes but also adapts to rapid changes in network structure, ensuring efficient resource allocation and robust optimization in dynamic environments. These findings underscore the potential of MLE-GA as a universal solution for complex social network problems where timely and accurate decision-making is imperative.