In the context of complex digital media ecosystems, effective dissemination of marketing content demands robust, adaptive, and structure-aware optimization strategies. This study proposes a novel framework—Complex Network Propagation with Divide-and-Conquer (CNP-Div)—designed to enhance information propagation in large-scale media networks by leveraging modular path optimization and probabilistic triplet modeling. The proposed model represents dissemination events as structured triplets (source, path, target) and formulates propagation likelihood as a function of source influence, platform channel weight, and audience receptivity. To address the computational difficulty of global optimization in dynamic media networks, a divide-and-conquer strategy is implemented: the global graph is partitioned into semantically coherent subgraphs, local path optimizations are executed within each, and a global aggregation phase ensures cross-community coherence. The framework further incorporates user feedback, platform centrality, and temporal factors to dynamically adjust propagation probabilities. Extensive experiments across simulated and real-world dissemination scenarios demonstrate that CNP-Div consistently outperforms baseline strategies—including random walk, greedy heuristics, and centrality-based models—across multiple metrics: propagation reach, average depth, engagement score, redundancy reduction, and runtime efficiency. Visualizations including heatmaps and precision–recall curves reveal that CNP-Div achieves more balanced content visibility across user activity phases and platform layers. Ablation results highlight the roles of triplet modeling, feedback reinforcement, and hierarchical path optimization in achieving these improvements.
