Cluster Analysis Using N-gram Statistics for Daihinmin Programs and Performance Evaluations

Abstract

In this paper, the authors elucidate the characteristics of the computer game Daihinmin, a popular Japanese card game that uses imperfect information. They first propose a method to extract feature values using n-gram statistics and a cluster analysis method that employs feature values. By representing the program card hands as several symbols, and the order of hands as simplified symbol strings, they obtain data that is suitable for feature extraction. The authors then evaluate the effectiveness of the proposed method through computer experiments. In these experiments, they apply their method to ten programs that were used in the UEC Computer Daihinmin Convention. In addition, the authors evaluate the robustness of the proposed method and apply it to recent programs. Finally, they show that their proposed method can successfully cluster Daihinmin programs with high probability.