Incremental association rule mining based on artificial immune

Most of the incremental association rule mining methods must rerun through processed data and have not made the best of the given rules. In this paper we propose an incremental association rules algorithm, this algorithm applies artificial immune theory and takes advantage of the given rules produced by original data set. Based on the quickly response during the memory cell recognizing the antigen, the algorithm is faster. The best rules are selected from the given rules as the optimum memory cell through promoting or inhibiting the antibodies, so the interest of the final rules is improved. The algorithm has better performance compared to FUP algorithm, especially in the number of the new transactions is less than half the number of transactions in the original data set, the time-consuming of the FUP algorithm is more than 10 times to this algorithm.