通过引入聚类竞争机制,提出了一种基于免疫聚类竞争的关联规则挖掘算法。将数据原始记录和候选模式分别作为抗原和识别抗体,通过聚类竞争加速克隆扩增,提高抗体成熟力及亲和性,增强候选模式支持度。实验及应用表明,该算法加快了关联规则挖掘的收敛速度,具有更强的全局与局部搜索能力,提高了所得关联规则的准确率。
By introducing a mechanism of Cluster and Competition,this paper proposes a novel Association rule Mining Algorithm based Immune Cluster and Competition. Raw datas are regarded as antigen and candidate patterns are regarded as antibody. Through the antibody clustering and compete, enhances the antibody's affinity maturation rate and improves the support of candi- date patterns. The simulation and real application illustrate that this algorithm can increase the convergence velocity and advance veracity of the association rule,and has the remarkable quality of the global and local research reliability.