基于TrueSkill模型的围棋棋手排名方法及评估  

A Ranking Approach of Go Players Based on TrueSkill Model with Empirical Evaluation

作　　者：吴霖[1] 陈磊[1] 邓超[1] 袁梅宇[1] 江虹[1] 

机构地区：[1]昆明理工大学信息工程与自动化学院,云南昆明650500

出　　处：《昆明理工大学学报（自然科学版）》2013年第3期47-55,共9页Journal of Kunming University of Science and Technology(Natural Science)

基　　金：国家自然科学基金(61163004);云南省应用基础研究面上项目(2010CD027)

摘　　要：如何根据比赛结果自动地估计参赛选手的真实竞技水平,是现代体育排名算法的核心研究内容.本文在综合分析现有等级分制、Elo算法和TrueSkill算法的基础上,通过仿真数据和真实数据,对比分析了TrueSkill排名、Elo排名和棋手段位间的差异及其原因.并且从排名的客观性、时效性、稳定性以及准确性四个方面,对TrueSkill算法的排名结果进行实际应用的可行性分析.实验结果表明TrueSkill的排名结果有较好的客观性、时效性、稳定性以及准确性,在仿真数据中TrueSkill的排名比Elo更合理.How to estimate player skill levels from game results is a key part of modern ranking algorithms for sports events. This paper firstly compares and analyzes Elo algorithm, TrueSkill algorithm, and the existing grad- ing system used in Go community. Simulation data and actual game results are then employed to generate ranking results according to these methods. Possible reasons for the differences are also offered. A feasibility analysis of using TrueSkill algorithm in real world practice is carried out by using objectivity, respond delay, stability, and accuracy as the rating criteria. Experiments show that rankings generated by TrueSkill algorithm have good prop- erties and it outperforms Elo in simulations by giving more sensible rankings.

关 键 词：评分系统 围棋 贝叶斯 TrueSkill 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]