Social Media‐Driven Emergency Decision Framework With Probabilistic Linguistic and Dynamic Consensus Approach

Emergency decision‐making requires rapid responses under high uncertainty and incomplete information. While traditional systems rely on hierarchical communication structures with inherent delays, social media provides real‐time situational awareness that remains difficult to integrate due to noise, misinformation and linguistic ambiguity. To address this gap, this study proposes a hybrid social media‐driven emergency decision framework that systematically integrates unstructured data into formal group decision‐making. The novelty of the framework lies in three aspects: (1) a hybrid LDA–TF‐IDF information fusion mechanism incorporating urgency and credibility weighting for robust attribute extraction; (2) a dual‐layer uncertainty modelling approach combining probabilistic linguistic term sets (PLTS) and interval‐valued distributions (IVDs) to capture both linguistic ambiguity and inter‐group preference dispersion and (3) a dynamic consensus mechanism with distance‐based adaptive feedback to improve convergence efficiency while preserving stakeholder heterogeneity. Experiments on three real‐world emergency events demonstrate that the proposed method outperforms benchmark approaches, achieving a Kendall's of 0.847 and Spearman's of 0.891, with a stability index of 0.923. The model reaches consensus within 3 iterations on average, improving convergence efficiency compared to traditional methods. Ablation studies confirm the effectiveness of each module, while sensitivity analysis verifies robustness across parameter variations. The proposed framework provides a scalable and interpretable solution for real‐time multi‐stakeholder emergency decision‐making, effectively bridging the gap between social media analytics and structured decision models.