WIDE:海量数据的聚类算法

给出了一种新的处理海量数据的聚类算法WIDE（window-density clustering algorithm）．它通过网格方法将数据之间的相互关联局部化，通过窗口技术来提高算法的效率，通过密度方法提高聚类的精度．以窗口为中介将网格方法和密度方法融合在一起是算法的主要思想．在此基础上对算法进行了扩展，在功能方面实现了混合型数据聚类、含障碍物数据聚类和增量数据聚类；在速度方面实现了分布式并行聚类．WIDE算法能够在局域网中的多台计算机上并行工作，效率高，计算复杂度为O（N），且能够发现任意形状的聚类，对噪声不敏感．

WIDE: Clustering Algorithm for Very Large Databases

A new clustering algorithm WIDE(window-density clustering algorithm)was introduced for very large databases.The relevance of data is localized through grid method.It uses window method to obtain high efficiency and density method to obtain high accuracy.The main idea of WIDE is to integrate gird method and density method with window technique,which makes WIDE powerful.Based on it,some extensions were made.Three new functions are added:it can deal with data of mixed types;it can do clustering in a data space containing obstacle objects;it introduces incremental clustering method for new coming data in an already pro- cessed database.In order to speed up clustering,it realizes distribution parallel clustering that can be imple- mented on a number of workstations connected via local area network.The advantages are:it is very efficient with a complexity of O(N);it is effective in discovering clusters of arbitrary shape;it is not sensitive to noise.