A Comprehensive Framework for Identifying Key Rumor Refuters for Combating Online Rumors

The widespread dissemination of rumors on social media presents a significant challenge, particularly in identifying users who actively and effectively counter misinformation. Existing approaches often rely heavily on engagement metrics, which can misrepresent influence by overlooking users who are structurally central and consistently opposed to falsehoods. This study introduces a holistic framework that integrates social network analysis (SNA) with multicriteria decision-making (MCDM) and sentiment analyses to identify and rank key rumor refuters to control the spread of rumors. Using real-world Bengali rumor data from Facebook, user interactions are modeled as a weighted undirected network, where edge weights reflect reaction types. Sentiment polarity, extracted from a fine-tuned BanglaBERT model, is incorporated as an independent ranking criterion rather than as an edge-weight modifier. Centrality, sentiment, and engagement scores are objectively weighted using CRiteria Importance Through Intercriteria Correlation (CRITIC) and then combined via Technique for Order Preference by Similarity to Ideal Solution (TOPSIS), with a sensitivity analysis conducted to assess the robustness of the rankings. The results show that the most significant refuters are those who combine sustained engagement with structurally central positions, while oppositional sentiment functions as a smaller but discriminating criterion that separates otherwise comparable candidates. This underscores the need for a multifaceted strategy to identify effective refuters, offering actionable insights for researchers, fact-checkers, and platform moderators working to combat misinformation.