双向关联规则挖掘及其相关性分析  

Analysis of correlation based on bidirectional association rules

作　　者：曾令明[1] 唐常杰[1] 阴小雄[1] 李川[1] 胡建军[1] 蒋永光[2] 

机构地区：[1]四川大学计算机学院 [2]成都中医药大学药学院

出　　处：《计算机工程与设计》2005年第10期2585-2588,2633,共5页Computer Engineering and Design

基　　金：国家自然科学基金项目(60073046);教育部高校博士学科点基金项目(20020610007)

摘　　要：分析了关联规则挖掘存在的3个问题:①不能有效地发现低支持度高置信度的有趣规则;②不能确定“相互依赖”的规则,如中药方剂中的药对药组;③找到的规则并不一定是有趣的,甚至是错误的。鉴于此,提出了双向关联规则挖掘,并对其进行了相关性分析。最后,将新方法用于中药方剂配伍规律的研究中,发现了大量具有重要意义的药对药组,证明新方法能很好地解决上述问题。Three problems in the association rules are analysed： ① The interesting rules that have low support but high confidence cannot effectively be found out; ② The interdependent rules, such as Traditional Chinese Medicine couples and groups cannot be discovered; ③ The findings are not always interesting rules, even are wrong, Thus, the bidirectional association rule and its correlation were analyzed based on the bidirectional association rule. Last, the new method was used in the study of the compatibility of Tranditional Chinese Medicine （TCM） and found out many significant TCM couples and groups. It is proved the method is able to solve above problems well.

关 键 词：数据挖掘 双向关联规则 相关性 药对 药组 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]