ArgRE: Formal Argumentation for Conflict Resolution in Multi-Agent Requirements Negotiation

ArgRE introduces a formal approach to conflict resolution in multi-agent requirements negotiation by embedding Dung-style abstract argumentation. In this system, each proposal, critique, and refinement is explicitly modeled as an argument, with conflicts represented as directed attack relations. The accepted set of arguments is computed under grounded and preferred semantics, providing a transparent and auditable decision-making process. The pipeline also integrates KAOS goal modeling, multi-layer verification, and standards-oriented artifact generation.

Evaluated across five case studies in safety-critical, financial, and information-system domains, ArgRE demonstrated significant improvements over heuristic synthesis. Independent evaluators rated its decision justifications significantly higher (4.32 vs. 3.07, p < 0.001). Semantic intent preservation remained comparable at 94.9% BERTScore F1, while compliance coverage reached 84.7% versus 47.6%–47.8% for baselines. Structural analysis confirmed that the default pairwise protocol yields acyclic graphs where grounded and preferred semantics coincide, whereas cross-pair arbitration introduces controlled cyclicity with predictable divergence.