Utilization of the preceding evaluation based on KPV Method

The KPV Method is an evaluation method of candidate moves. The method was developed for a knowledge based game program while one of the authors has presented. In this paper, it is shown that the evaluation of former candidates can be utilized to evaluate the present candidates in a game program based on the KPV Method. After introducing the KPV Method, a cut algorithm, called NV-cut, is presented. Examples of an Othello game show that the algorithm can cut unnecessary candidates by using knowledge named ‘necessity decision knowledge.’ The cut ratio increases when a play program also utilizes the opponent’s thinking time. The authors also discuss the commonly associated errors and a way to avoid them. Finally then discuss the basis of the NV-cut. [In Japanese.]