Abstract
The research on Turn-based strategy games(TBSG) has not been well conducted as Shogi and Go, and computer AIs of TBSG are usually weaker than human players. There are many choices in one turn because of the combination of the order to move units and each actions. It is possible to reduce divergences per node by creating the game tree divided per actions. The number of candidate choices remains big even if divergences per node are reduced, but the search can become more effective by removing bad choices from the tree. This study applies fuzzy evaluation to remove bad choices by decision based on human intuition.
