🤖 AI Summary
This study addresses the absence of a formal modeling framework in strategic crisis analysis that operates without requiring complete payoff or probabilistic information, integrates expert qualitative judgments, explicitly captures dependencies, and supports auditable update rules. To bridge this gap, the authors propose a two-layer formal framework that decouples a static scenario database from a dynamic scenario tree system. They introduce, for the first time, a formally defined extended scenario bundle analysis model, incorporating a domain-modifier layer, a topological structure over scenario space, a typed state-update mechanism, and a multi-criteria evaluation method. This architecture enables context-sensitive modeling of multi-agent attitudes—including beliefs, desires, intentions, fears, and coalition commitments—while maintaining mathematical rigor and computational tractability, thereby significantly enhancing the transparency, traceability, and expressive power of complex crisis analysis.
📝 Abstract
Strategic crisis analysis needs representations that combine qualitative expert judgement, explicit interdependence, and auditable update rules without requiring fully specified payoffs or probabilities. Scenario Bundle Analysis (SBA), developed by Amos Perlmutter and Reinhard Selten, provides such a starting point, but the original formulation leaves several database, topology, and update interfaces implicit. This paper presents a formal refinement and extension of the original SBA framework, introducing a two-layer architecture that separates a static scenario database from a dynamic scenario tree system. The extended framework incorporates a richer attitude vocabulary: beliefs, desires, intentions, fears, and coalitional commitments, with expectations treated as doxastic attitudes. It also adds a domain/modifier layer for contextual framing, a topology on admissible scenario spaces\index{Scenario space}, typed assessment-state updates, and multi-criteria evaluation. Mathematical definitions are stated with sufficient precision to support computational implementation.