A Unified Framework for Fake News Detection using Content Community and Propagation Features

Social media means pose a serious threat to social trust, democracy, and the discourse that occurs in the society through various efforts to broadcast fake news. Traditional fake news detection systems often rely solely on content-based features, making them highly susceptible to manipulation and adversarial attacks. This paper introduces a novel community-sensitive framework for fake news detection that integrates social network analysis, user behavior modeling, and misinformation propagation patterns. By incorporating multi-dimensional features, our framework captures crucial signals—such as community influence and diffusion dynamics—that are often missed by purely textual approaches. Experimental results demonstrate that leveraging community structures and the peculiarities of misinformation spread significantly enhances detection accuracy and stability compared to content-only baseline models on benchmark datasets.