Designing a high-performance Chinese chess computer player for middle game position

Computer game is an important experimental plot for testing the performances of artificial intelligence algorithms. In comparison with the international chess computer game, Chinese chess computer game (CCCG) has more complicated and intricate variations of feasible moves. Particularly, due to the complexity of positions and limitation of computational time, the middle game position plays a very important role when developing an intelligent CCCG player. In this paper, a high-performance CCCG player named CCCG_ylh for middle game position is designed and implemented by 1) designing a more flexible data strncture to describe the board position, pieces, and legal moves, in which the board and pieces arrays are mutually dependent; 2) using the alpha-beta algorithm to search a pruned game tree which can be formed by only storing the pieces array; and 3) constructing a simplified evaluation function which can appraise a position with more little time consumption. It will make the alpha-beta algorithm search the game tree more deeply so that more accurate move can be selected. By comparing with the testing players, the better performance of CCCG_ylh is demonstrated. The experimental results show that CCCG_ylh can obtain the higher playing level in the middle game positions.