近年來電腦硬體效能的增長,尤其在運算能力與儲存空間的部分,已有顯著的發展。如何利用電腦的效能以提升電腦對局程式的能力,就成了一個有趣的議題。 由於電腦的優勢在於快速的計算與大量的儲存空間,所以可以利用此一優勢來讓電腦來模擬人類的下棋時的記憶與推理。透過大量的棋譜記憶,就如同人類下棋的經驗學習,可以省去電腦思考的時間,而達到相當於專家水準的能力。 本論文利用知識庫的架構,大量蒐集對局資訊,建構具備學習能力的Othello對局系統。藉由知識庫中的大量對局資訊,系統對局時可由知識庫分析著手的優劣,從而得知最佳的著手。且從對局中累積資訊加入對局系統知識庫中,使得棋力可以自我學習成長。 本論文以4x4 Othello與6x6 Othello做為實驗的對象,設計對局系統與知識庫架構,並有效地提升整體對局系統的棋力。
In recent years, due to the tremendous development of computer hardware, the computation power and storage space have grown rapidly. How to utilize the efficiency of the computer to promote the competence level of a game playing program has become an interesting issue. Owing to the fast calculation and large storage ability of computers, we enable the computers to simulate the memory and inference process of human beings when they are playing board games. It is possible to save so much computation time that the program may attain the level of experts via collecting and analyzing numerous game playing records. In this paper, we designed and set up an Othello program system with powerful capability of learning by collecting and analyzing information from game playing records. Based on such a knowledge-base, the system was able to analyze the pro and con of each move and select the best one. We conducted experiments on four-by-four Othello and six-by-six Othello. The competence level of our program can be enhanced by accumulating information from game playing records into the knowledge-base from time to time.