Graph model for conflict resolution of large-scale group with overlapping coalitions based on meta-clustering under big data

Subjectivity in classical Graph Model for Conflict Resolution (GMCR) modeling remains a primary concern and scales poorly to large-scale group (LSG) conflicts on social media. In LSG conflicts, (i) difficulties in obtaining and measuring conflict data excessively restrict model objectivity and applications, and (ii) overlapping coalitions create internal vetoes that standard reachability ignores. This motivation calls for a data-driven GMCR that can discover stakeholder groups, quantify overlaps, and preserve interpretability. Accordingly, this research starts from a data perspective and proposes a novel LSG-GMCR model with overlapping coalitions to obtain conflict equilibrium solutions. Considering the requirements of massive data with potential attributes, we employ a crawler algorithm to extract conflict texts from posts, comments, and comment replies on Weibo. A triple conflict data preprocessing mechanism is designed to accurately tackle the scattered forms and large scales. Multiple technologies are integrated to extract features. Then, a meta-clustering algorithm proposed in this paper is constructed for features with different attributes to classify clusters as the group decision makers. A fuzzy membership matrix is also yielded for overlap computation. Finally, the LSG-GMCR model with overlapping is established, supported by the formalized directed overlap and admissibility filter. For validation and verification purposes, this proposed framework is applied to investigate actual conflicts with respect to controversial issues like power rationing conflicts. The approach enhances the scalability, automation, and realism of OR-based conflict analysis and is directly applicable to policy design in complex LSG settings.