Abstract:

Association rules mining is an important problem in data mining. The traditional mining algorithms have high complexity and low efficiency, while the intelligent algorithms have the advantages of maintenance of population diversity and robustness in the searching process. A model of mining association rules based on immune clone culture algorithm is proposed. This model takes advantages of global searching in the immune clone algorithm to rapidly search the frequent item sets and then extract the interesting rules. It also uses the knowledge structure of belief space in the culture algorithm to guide the population’s evolution and enhance the purpose and directivity of searching. The experiments show that the new model has faster performance speed and also improves the accuracy of the rules.

Key words: association rules;immune clone algorithm;culture algorithm