摘要
研究多关系数据挖掘的聚类问题,提出一种有效的多关系聚类算法EMC.EMC算法的目标是提高聚类的准确率,并且降低运行时间.EMC算法首先利用元组ID传播的思想,计算两个对象之间的相似度,接着利用K中心点聚类算法,将对象划分成簇.实验表明,EMC算法显著降低运行时间,并且提高聚类的准确率.
The problem of clustering in multi-relation data mining was investigated, and an efficient multi-relational clustering algorithm called EMC was proposed. EMC aims at increasing the accuracy of clustering, and decreasing running time. First, EMC computed the similarity between two objects by taking advantage of tuple ID propagation approach. Then, EMC clustered the objects by K-medoids clustering algorithm. Performance results demonstrate that, EMC significantly decreases running time, and increases the accuracy of clustering.
出处
《微电子学与计算机》
CSCD
北大核心
2016年第4期133-137,共5页
Microelectronics & Computer
基金
国家自然科学基金(61462008
61472254
61420106010)
关键词
多关系数据挖掘
聚类
元组ID传播
相似度
K中心点聚类算法
multi-relational data mining
clustering
tuple ID propagation
similarity
K-medoids clustering algo- rithm