Feature Extraction and Cluster Analysis Using N-gram Statistics for DAIHINMIN Programs

In this paper, we elucidate the characteristics of the computer program to play DAIHINMIN, which is a popular Japanese card game with imperfect information. First, we propose a method to extract feature values using the n-gram statistics and a cluster analysis method using the feature values. By representing the program hands as several symbols and representing the order of hands as simplified symbol strings, we obtained the data suitable for feature extraction. Next, we evaluated the effectiveness of the proposed method through computer experiments. In these experiments, we applied our method to ten programs which were used in UECda contest. Finally, we show that our proposed method can successfully cluster DAIHINMIN programs with high probability.