Abstract: Business activity and engineering practice always accumulate large dataset with important information.But because of the dataset’s largeness and frequent updating，if the Apriori based algorithm is applied to incremental association rules mining，it is not only inefficient，but also many redundant rules will be produced with low minimum support while some interesting rules will be lost with high minimum support，which leads to the algorithm’s blunt perception to those rules.So，following genetic principle，and combining with natural immune involution theory and relative bionic mechanism，this paper proposes an IOGA（Immune Optimization based Genetic Algorithm）approach for incremental association rules mining.Experiment demonstrates the proposed method’s effectiveness and presents its good performance in perceiving rules’ subtle change，reducing redundant rules and finding interesting rules.

Key words: immue optimization, genetic algorithm, association rules, incremental mining

摘要： 商业活动和工程实践中通常会积累一些大规模的携带重要信息的数据，由于这种数据集经常有更新且数据量较大，在对它们进行增量式关联规则挖掘时，若采用基于传统的Apriori算法进行计算，一方面难以取得较好的效率；另一方面支持度设置过低会产生大量的冗余规则，设置过高则会把一些支持度不高但有用的规则过滤掉而导致算法对这些新规则感应迟钝。因此，借助遗传算法的相关机理，同时结合自然界的免疫进化理论及相关仿生机制，提出一种IOGA（Immune Optimization based Genetic Algorithm，基于免疫优化的遗传算法）增量式关联规则挖掘方法。通过实验表明，该方法应用于大规模数据集的增量式关联规则挖掘时，可以及时地感知规则的变更并发现有用的规则，减少了冗余规则的产生，同时挖掘效率也有明显提高。

关键词: 免疫优化, 遗传算法, 关联规则, 增量式挖掘