一种基于聚类的多数据库分类方法设计

Design of classification for multi-database mining based on clustering method

多数据库挖掘最常用的方法是先将多数据库进行分类,然后对每个类进行单独挖掘,最后将各个类中的模式进行集成得到全局模式。这些数据库分类方法都只是针对事务数据库而设计,用两个数据库中共同项集的比例来衡量这两个数据库的相似度,以此来进行数据库分类。本文提出一种基于聚类的数据库分类方法,可以对任何类型的数据库进行分类。

The common method for multi-database mining is to classify all the databases first,and then mine each class separately,after that,the local patterns in all classes are integrated to be a global pattern.This kind of classification methods are only designed for the transaction databases,using their common itemsets to measure their similarity.A new database classification method based on clustering methodlogy was proposed in this paper,which adapts to different data features including itemset and can be used in any other types of database classification.