End-to-End Framework for Intelligent Strategic Agents in Competitive Games

Competitive multi-agent environments require intelligent agents to make decisions under uncertainty and evolving conditions. This paper presents a framework for strategic decision support that combines knowledge-based modeling, argumentative deductive reasoning and blockchain traceability within a modular architecture. The approach enriches game events by means of semantic annotations and represents strategies and contextual information in a Bipolar Weighted Argumentation Framework, enabling explainable strategy selection. Long-term logging with data integrity and decision traceability is supported by a dedicated notarization layer. A case study in a popular social deduction online game validates the framework in a dynamic setting with incomplete information. Preliminary experiments conducted on a containerized testbed provide initial evidence of its feasibility by assessing the resource usage and execution times of the core services.