Cycle Spectral Sequence Synthesis Models Misinformation Cascade Evolution Through Graph Dynamics and Cascade Cycles Lewis Foster, Nathan Bryant, Florence Blanc TUM School of Engineering and Design, Technical University of Munich, Munich, Germany Abstract

Abstract

The rapid dissemination of misinformation across digital social platforms presents a profound challenge to modern information ecosystems. Traditional approaches to modeling information cascades often rely on static network snapshots or continuous-time point processes that struggle to capture the complex, higher-order structural dynamics of propagation networks over time. To address these critical limitations, this paper introduces a novel methodological framework termed Cycle-Spectral Sequence Synthesis for modeling the evolution graphs of misinformation cascades. By conceptualizing the spreading process as a sequence of discrete but topologically contiguous graph states, the proposed framework leverages spectral graph theory to extract invariant features within the frequency domain. Specifically, we investigate the cycle-spectral properties of cascade networks, which encapsulate both the localized cyclic feedback loops of user interactions and the global structural frequencies of information flow. Furthermore, a sequence synthesis mechanism is developed to dynamically reconstruct missing or sparsely sampled propagation stages, ensuring a continuous and high-fidelity representation of the cascade trajectory. Comprehensive theoretical analysis and extensive evaluations on large-scale social network datasets demonstrate that this approach significantly outperforms existing baseline models in early misinformation detection and cascade size prediction. The findings provide a robust foundation for understanding the latent topological signatures of deceptive information diffusion and offer actionable insights for developing more resilient digital communication platforms.