从抽取到绪合:代理增强战略推理中的纠缠启发式方法
人工智能
2025-07-21 v1
摘要
我们提出了一种混合架构用于代理增强的战略推理,结合了启发式抽取、语义激活和组合绪合。我们的模型涵盖从古典军事理论到当代企业战略的各种来源,通过受量子认知研究启发的语义相互依存过程激活和组合多个启发式。不同于选择最佳规则的传统决策引擎,我们的系统将冲突的启发式融合为连贯且情境敏感的叙事,由语义交互建模和修辞框架引导。我们通过Meta与FTC案例研究演示了该框架,并通过语义指标进行初步验证。讨论了局限性和扩展(例如动态干扰调谐)。
引用
@article{arxiv.2507.13768,
title = {From Extraction to Synthesis: Entangled Heuristics for Agent-Augmented Strategic Reasoning},
author = {Renato Ghisellini and Remo Pareschi and Marco Pedroni and Giovanni Battista Raggi},
journal= {arXiv preprint arXiv:2507.13768},
year = {2025}
}
备注
Peer-reviewed full paper accepted through a double-blind review process at the HAR 2025 conference (https://har-conf.eu/). The official version will appear in a volume of the Lecture Notes in Computer Science (LNCS) series