Mining Association Rules with Constraints Based on Immune Genetic Algorithm

This paper proposed an algorithm of mining association rules with constraints based on immune genetic algorithm, which for the anti-monotone and monotone constraint conditions. This algorithm was inspired by the fundamentals of the biological immune system which the process of B-cells to produce the optimal antibody under the T-cells constraint. From this metaphor, firstly, according the different of the constrained condition the algorithm divided the individuals into two sets. Secondly, the algorithm was effortless to locate the optimal solution. Namely, it made the infeasible solutions along the direction of reduce constraint search the whole solution space and the feasible solutions rely on improving their fitness values locate the optimal solution. The time on discovered rules were compared with the traditional method and experimental results show that the effectiveness of the proposed algorithm.