基于阈值估计的强相关项目对挖掘算法  

Mining Algorithm of Strongly Correlated Item Pairs Based on Threshold Estimation

作　　者：李强[1] 张勇实[1] 

机构地区：[1]哈尔滨工程大学计算机科学与技术学院,哈尔滨150001

出　　处：《计算机工程》2010年第8期58-59,63,共3页Computer Engineering

基　　金：国家自然科学基金资助项目(60873037);黑龙江省自然科学基金资助项目(F2005-02)

摘　　要：将关系数据库中基于最小相关阈值的挖掘问题转为Top-K强相关项目对的挖掘,利用关系数据库的结构信息,有效地估计合适的阈值,提出基于阈值估计的Top-K强相关项目对挖掘算法。借助定理证明的形式在理论上推导该算法,并基于自主开发的仿真平台和权威数据库样本进行仿真实验。该算法能高效、快速地得到挖掘结果。The mining scheme based on minimum correlation threshold in relational database is transformed to the mining scheme on Top-K strongly correlated item pairs.By using structural information of relational database,appropriate threshold is effectively estimated,and an algorithm for Top-K mining strongly correlated item pairs based on evaluation of threshold is proposed.The theory of this algorithm is deduced with the help of the process of proving some theorems.Based on a simulation platform developed independently and two authorized samples of database,some experiments are conducted.The theoretically analysis and some simulations results show that this algorithm can obtain mining results efficiently and quickly.

关 键 词：数据挖掘 阈值 强相关项目对 关系数据库 皮尔森关联系数 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]