A Novel Explainable Rumor Detection Model with Fusing Objective Information

Amidst the dynamic expansion of social networks, the dissemination of rumors has accelerated, rendering rumor detection an imperative and formidable endeavor in the realm of online environment governance. Traditional rumor detection methodologies have predominantly neglected the significance of interpretability. To rectify this deficiency, we introduce a sophisticated and interpretable rumor detection model, denoted as FOEGCN. This avant-garde model discerns objective information from an extensive database predicated on subjective data, subsequently employing a graph neural network to classify rumors based on a fusion of objective and subjective intelligence. Concurrently, FOEGCN elucidates the detection results via a visually compelling interpretation. Rigorous experiments conducted on a pair of publicly accessible datasets substantiate that our proposed model surpasses existing baseline methods in both rumor and early rumor detection assignments. The FOEGCN model enhances performance by 1% and 1.6% in terms of accuracy metrics. A comprehensive case study further accentuates the model’s superior interpretability, making it an exemplary solution for tackling the challenges of rumor detection.