基于免疫遗传算法的多维多层关联规则挖掘  

Multi-dimension and Multi-level Association Rule Mining Based on Immune Genetic Algorithm

作　　者：朱玉[1] 张虹[1] 孔令东[1] 

机构地区：[1]中国矿业大学环境与测绘学院,徐州221008

出　　处：《计算机工程》2009年第23期181-183,186,共4页Computer Engineering

基　　金：江苏省自然科学基金资助项目(BK2005021);江苏省普通高校研究生科研创新计划基金资助项目(CXB-1392)

摘　　要：提出一种基于免疫遗传算法的多维多层关联规则挖掘算法。免疫遗传算法具有很好的鲁棒性及全局搜索能力,能快速有效地进行全局优化搜索。针对现有多维多层关联规则挖掘中存在阈值定义不合理的缺点,依据多维和多层数据的共同特点,给出一种启发式的阈值自定义方式,结合免疫遗传算法提高挖掘效率和结果的准确性。结果表明,挖掘效率和质量有明显提高。This paper proposes a method of mining multi-dimension and multi-level association rule based on the Immune Genetic Algorithm（IGA）. The algorithm has good robustness and good whole global search capability. It searches more quickly and efficiently in the whole global optimization. It puts forward a heuristic user-defined threshold method, which based on the common characteristic of multi-dimensional and multi-level data, to overcome the drawbacks caused by the unreasonable method of defining threshold. Combined with the immune genetic algorithm, the precision and efficiency of mining association rules is improved. Results show that the efficiency and quality improved obviously.

关 键 词：免疫遗传算法 多维多层 关联规则 数据挖掘 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]