Rumor Detection Integrating Dynamic Communication Structures and Semantic Enhancement

Abstract

In the contemporary era of information proliferation, the viral spread of online rumors poses a critical challenge to social governance. Conventional detection methodologies predominantly utilize static analysis of textual and visual modalities, which fails to account for
the dynamic evolutionary characteristics of misinformation. Consequently, these models often lack the robustness required for the timely detection of rapidly propagating rumors.
??This paper proposes a paradigm shift in rumor detection, transitioning from content-centric analysis to a propagation-oriented framework.
By modeling the rumor dissemination process as a dynamic spatiotemporal sequence, we develop a deep learning architecture designed to
capture high-dimensional structural features. Unlike existing approaches limited by local or static perspectives, our model effectively captures
long-range structural dependencies and global temporal dynamics across the entire propagation lifecycle. Empirical evaluations on benchmark
datasets demonstrate that the proposed model significantly improves detection latency and accuracy, offering a robust theoretical and technical
foundation for automated public opinion monitoring and proactive intervention systems.