Characteristics and Governance Pathways of Online Rumor Diffusion in the New Media Environment: An Empirical Analysis Based on Weibo and Multi-Source Fact-Checking Data

Abstract

This study performs a computational communication analysis grounded in communication studies and computational social science. The study integrates four datasets—CED, CHECKED, ChineseRumor, and MCFEND—covering a total of 61,134 root posts and 300,000 interaction texts. It proposes a three-dimensional analytical framework of emotional communication that integrates—‘text, emotion, topic shift, and a diffusion network.’ In this study, sentiment gap (ΔS), topic shift (ΔT), and diffusion depth and intermediary position are treated as indicators of rumor diffusion. The results reveal (1) in the context of Weibo, rumor posts are more likely to have a larger average number of child posts as well as deeper levels of diffusion and stronger betweenness centrality in the context of key posts as compared to posts that are non-rumor in nature, (2) in a multi-source fact-checking platform, rumor samples are more likely to have a larger level of user engagement and show more complex diffusion patterns, and (3) most of the time, topic shift indicators remain high, indicating that most of the time, rumors are spread that are disconnected from the original context and are surrounded by a re-narrativization process. The study also proposes a collaborative governance pathway characterized by early monitoring, embedded verification, and tiered governance. The study aims to provide empirical evidence for rumor identification and governance in Chinese-language platform environments.