基于网格和密度的数据流聚类算法  

Data stream clustering algorithm based on grid and density

作　　者：张丽[1,2] 胡颖[2] 

机构地区：[1]河南大学计算机信息与工程学院,河南开封475004 [2]商丘师范学院计算机科学与技术系,河南商丘476000

出　　处：《郑州轻工业学院学报（自然科学版）》2010年第4期75-78,84,共5页Journal of Zhengzhou University of Light Industry:Natural Science

摘　　要：针对数据流的特点,提出了一种新的网格密度结合的GCTS算法.该算法采用双层架构,在线层实现了网格密度参数的自设定,离线层以网格单元的重心为中心点,建立一个最大的子网格,使候选网格中的局部密集区域转化成了密集网格.使用最小生成树的算法生成聚类结果,提高了聚类效果.According to the characteristics of the data stream,a new clustering algorithm GTCS which combined the approach based on density and grid was presented.By means of the model of double-layer construction,the method set the key of densities of the data grids automatically in online layer.The offline layer using the data gravity for the center,a maximum of subgrid is built.It makes the dense regions of the candidate grids into dense grid.It uses the minimum spanning tree clustering algorithm to get the clustering results and improve the clustering affect.

关 键 词：数据流 聚类算法 子网格 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]