DDGCN: Dual Dynamic Graph Convolutional Networks for Rumor Detection on Social Media

Abstract

Detecting rumors on social media has become particular important due to the rapid dissemination and adverse impacts on our lives. Though a set of rumor detection models have exploited the message propagation structural or temporal information, they seldom model them altogether to enjoy the best of both worlds. Moreover, the dynamics of knowledge information associated with the comments are not involved, either. To this end, we propose a novel Dual-Dynamic Graph Convolutional Networks, termed as DDGCN, which can model the dynamics of messages in propagation as well as the dynamics of the background knowledge from Knowledge graphs in one unified framework. Specifically, two Graph Convolutional Networks are adopted to capture the above two types of structure information at different time stages, which are then combined with a temporal fusing unit. This allows for learning the dynamic event representations in a more fine-grained manner, and incrementally aggregating them to capture the cascading effect for better rumor detection. Extensive experiments on two public real-world datasets demonstrate that our proposal yields significant improvements compared to strong baselines and can detect rumors at early stages.

DGTR: Dynamic graph transformer for rumor detection

“…DDGCN w/o knowledge (Sun et al, 2022 ): A dynamic graph-based rumor detection work. DDGCN simulates both rumor propagation and knowledge evolution during it.…”

“…Song et al (2021bSong et al ( , 2022 proposed modeling rumor propagation patterns using dynamic graphs, using GCN to encode structural information, gating networks to encode temporal information, and average pooling of individual node embeddings to produce the full graph representation. Sun et al (2022) leverage external knowledge to improve the model's comprehension of the text, while the way of encoding spatial and temporal information is similar with the previous methods. The GCN-based structural encoder layer can only collect first-order neighbor information, which is limited by perceptual field size and difficult to capture spatial long-range dependencies.…”

“…Similar to previous study (Song et al, 2022;Sun et al, 2022), the source tweet, retweets, and replies are considered nodes. We consider retweet or respond behaviors as edges.…”

Trust-Aware Evidence Reasoning and Spatiotemporal Feature Aggregation for Explainable Fake News Detection

“…• DDGCN [28]: DDGCN is a dynamic graph convolution neural network model to capture the characteristics of the information propagation structure and knowledge entity structure at each point in time. Since our model only concentrates on the contents and social contexts, we don't introduce dynamic knowledge structure.…”

“…Reference source not found., both comments T8 and T9 are replying to comment T6 with the same spatial structure characteristics, but from the perspective of time, there are clear differences between them (T8 is released earlier than T9, meaning users may already be affected by T8 when T9 is released). Recent studies [25][26][27][28] have demonstrated that temporal structure features can capture the dynamic evolution of information in a more finegrained manner and promote the early detection performance. Spatial and temporal structures depict the evolution of news messages from the perspectives of information interaction network and temporal message propagation respectively, which are complementary.…”