Temporal-Structural Propagation Graph Analysis for Coordinated Misinformation Campaign Detection and Source Attribution in Social Networks

Abstract

The proliferation of coordinated misinformation campaigns across large-scale social networks poses critical challenges to information ecosystem integrity and democratic discourse. Existing detection approaches predominantly rely on either content-based signals or simplified graph topology, limiting their capacity to capture the nuanced behavioral synchronization patterns that characterize organized influence operations. This paper presents a temporal-structural propagation graph analysis framework that jointly models the topological characteristics and dynamic spreading behaviors of information cascades to identify coordinated misinformation networks and trace their origin nodes. The proposed approach constructs heterogeneous propagation graphs from timestamped user interaction streams, extracts community-level synchronization features alongside cascade velocity anomaly indicators, and employs a multi-hop diffusion backtracking procedure that assigns calibrated confidence scores to candidate source nodes through structural centrality constraints. Experiments conducted on three publicly available benchmark datasets demonstrate consistent improvements over representative graph-based and hybrid detection baselines in both coordinated campaign identification and source localization accuracy. Ablation results confirm that temporal dynamics and structural topology operate as mutually reinforcing signal streams whose combination yields the strongest attribution outcomes. These findings carry direct implications for platform-level intervention design and national information security policy objectives.