一种基于免疫聚类竞争的关联规则挖掘算法  

Novel association rule mining algorithm based on immune cluster and competition

作　　者：徐雪松[1] 章兢[1] 贺庆[2] 

机构地区：[1]湖南大学电气与信息工程学院,长沙410082 [2]中南大学信息科学与工程学院,长沙410082

出　　处：《计算机工程与应用》2007年第16期16-19,29,共5页Computer Engineering and Applications

基　　金：国家自然科学基金重点项目(National Natural Science Foundation of China No.60634020);教育部博士点基金项目(the Ph.D. Programs Foundation of Ministry of Education of China No.20060532026)

摘　　要：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.

关 键 词：关联规则 聚类竞争 克隆选择 数据挖掘 

分 类 号：TP182[自动化与计算机技术—控制理论与控制工程]