Big2 is a card game that is very popular is east asia nad south east asia. strategy and probability to win this card game is very complex. compare to chess and other board games , research of artificial inteligence algorithm to be implemented in card games such as big2, in infree=quent . for that reason, this paper proposal an approach that can be utilized in solving card games problem, especially big2, bayesian network, minimax, and rule based system. based on the comparison of of those there candidates and charactericstic of big2 card games, combination of bayesian network and minimax algorithm is chosen to optimize the winning probability. the approach has been implemented in java platform. expriment shows that this approach has 23% higher probability compere ti average winning probability of human player and compare to the approach using greedy algorithm.