A knowledge based approach to planning in card games

The purpose of this research is to explore computational approaches to the planning in card games. Because of the complexity of the problem, the study was restricted to the Declarer playing in a NT-contract in Bridge. A program has been provided which simulates the planning process of a human player. It uses a knowledge based approach which encodes knowledge in form of rules and in form of plans. The program uses rules to select playing techniques for each suit held in the player's hand during static analysis and to guide a small tree search which confirms a particular technique is best. Once a technique is selected, the plans are used to construct the playing sequence of cards for this technique. The possibilities and limitations of the overall approach are discussed.