在多数据库中确定支持度的有效方法

Efficient strategy to obtain support in multi-database

提出了一种用于在多数据库环境下确定模式的支持度的方法。数据挖掘总是在强关联规则模式中进行挖掘,因此计算模式的支持度必不可少。由于多数据库与单数据库的本质不同,单数据库中的支持度只具有本地效应,它的数值和计算方法在多数据库中都是不适用的,因此需要找到一个在多数据库中计算支持度的方法。又由于每个本地数据库的各方面情况不同,因此各自在总部制定决策时的地位和重要性也就各异。于是,在考虑某模式的支持度时,不能简单的将该模式在各本地数据库中的支持度进行平均,而需要引进"权值"的概念。在综合考虑了这些因素之后,提出了一些确定权值的方案,并给出了一个基于权值计算支持度的方法。

A strategy is proposed, which is used to obtain the support of the pattem in multi-database. Since data mining always proceeds within strong association rules, calculating the support is necessary. Due to the difference between mono-database and multi-database, the support in mono-database only has local effect so that the strategy of calculating support and the support itself in mono-database is not proper in multi-database. So it is necessary to develop a new method. Beside, because of some aspects in all local databases is different, the status and importance of each local database in global-decision making is different. So, it is not proper to simply average all the support of each database respectively and the conception of weight is proposed. On considering the factors above, some methods are proposed to obtain the weight and offers an effective weight-based strategy to get the support.