Practical Strategic Reasoning for Intractable Games

棘手游戏的实用策略推理

• 批准号: 0414710
• 负责人: Michael Wellman
• 金额: --
• 依托单位: Regents of the University of Michigan - Ann Arbor
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2005
• 资助国家: 美国
• 起止时间: 2005-03-15 至 2008-12-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0414710&HistoricalAwards=false
• 关键词: Practical; Strategic; Reasoning; Intractable; Games

Many digitally-mediated activities present participants with complex strategic decisions. Strategic interaction in electronic commerce arises, for instance, in negotiation, trading, resource finding, matchmaking, advertising, recommendation, contracting, and executing. Scenarios in these applications are properly modeled as games, however, direct application of game theory quickly becomes infeasible as problem complexity grows. Despite enormous advances in game theory and mechanism design, an ever-growing number of applications are intractable using standard methods. Thus, designers must resort to highly stylized models, or non-strategic analysis using simulation or experimentation methods.This project bridges the gap between simulation and game-theoretic analysis by developing computational tools for empirical game-theoretic analysis. The new techniques combine gaming, high-fidelity simulation, experimental manipulation, statistical analysis, search, and game-theoretic reasoning in a principled way, to reveal the shape of complex strategic environments. The results of the project will have immediate practical application to strategic reasoning in many domains, including government auctions for spectrum, oil and gas, etc.; complex matching programs (e.g., medical residency); deregulated energy market operation; and business-to-business electronic commerce. By extending the scope of problems amenable to strategic analysis, this research will inform the design of more efficient resource allocation mechanisms for public as well as private-sector applications.

许多以数字为媒介的活动为参与者提供了复杂的战略决策。 例如，电子商务中的战略互动出现在谈判、交易、资源寻找、配对、广告、推荐、签约和执行中。 在这些应用程序中的场景被正确地建模为游戏，然而，随着问题复杂性的增加，直接应用博弈论很快变得不可行。尽管博弈论和机制设计取得了巨大的进步，但越来越多的应用程序使用标准方法是难以处理的。 因此，设计师必须采用高度程式化的模型，或者使用模拟或实验方法进行非战略性分析。本项目通过开发用于经验博弈分析的计算工具，弥合模拟和博弈分析之间的差距。 这些新技术将联合收割机博弈、高保真模拟、实验操作、统计分析、搜索和博弈论推理原则性地结合起来，以揭示复杂战略环境的形态。该项目的结果将立即实际应用于许多领域的战略推理，包括政府对频谱、石油和天然气等的拍卖;复杂的匹配程序（例如，医疗居住）;放松管制的能源市场运作;以及企业对企业的电子商务。通过扩大可进行战略分析的问题的范围，这项研究将为公共和私营部门应用程序设计更有效的资源分配机制提供信息。