基于OpenMP的Gibbs抽样正负关联规则算法研究  

Research on the Mining Algorithm of Gibbs Sampling Based on OpenMP

作　　者：徐佳 XU Jia(School of Information,Guizhou University of Finance and Economics,Guiyang Guizhou 550000,China)

机构地区：[1]贵州财经大学信息学院,贵州贵阳550000

出　　处：《信息与电脑》2022年第24期69-71,共3页Information & Computer

摘　　要：大数据集中挖掘正负关联规则是关联规则挖掘的重要研究内容。负关联规则挖掘存在挖掘关联规则数量多、难度大等问题,因此针对大数据集中挖掘正负关联规则提出一种基于OpenMP的Gibbs抽样正负关联规则挖掘算法。该算法通过Gibbs抽样从原始数据集中挖掘得到重要的关联规则,并在Gibbs抽样的转移概率计算部分利用OpenMP并行技术进行加速。在只挖掘重要正负关联规则的同时,缩短挖掘时间,有效提高正负关联规则挖掘的效率。在UCI蘑菇数据集中使用该算法,实验结果显示该算法在大数据集中具有较好的表现。Mining positive and negative association rules in big dataset is an important research content of association rules mining.There are many problems in mining negative association rules, such as large number and difficulty. This paper proposes a mining algorithm of Gibbs sampling for positive and negative association rules based on OpenMP. The algorithm mines important association rules from the original data set through Gibbs sampling, and uses OpenMP parallel technology to accelerate the calculation of Gibbs sampling transition probability. While mining important positive and negative association rules, it can reduce the mining time and effectively improve the efficiency of mining positive and negative association rules. The experimental results of using this algorithm in UCI mushroom data set show that this algorithm has a good performance.

关 键 词：GIBBS抽样 正负关联规则 OPENMP 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论]