基于属性值分类的特征规则的知识发现

The KDD of Characteristic Rules Based on Attribute Values Classification

1 引言 知识发现(Knowledge discovery in database,KDD)是应用一系列技术从大型数据库或数据仓库中提取隐含的、未知的、非平凡的对决策有潜在应用价值的知识和信息的过程,提取的知识表示为概念、规则、规律、模式等形式.知识发现过程一般由三个主要的阶段组成:数据准备,数据开采,结果表达和解释[2].

This paper presents a new classification technique of data mining and knowledge discovery in any scale relational database . Based on set theory , it divides relational table into several equal value categorys based on attribute values . calculates information capacity in decision factor of the every condition attribution , eliminates surplus attributions, and erases repeat units. Then it gets strong equal value categorys of the relational table being reduced data , gets strong characteristics on the each strong equal value category, and gathers characteristic knowledge rules: It overcomes the redundancy nature, complicated nature and unfit nature to big capacity data of some classification technique at present. It is better efficiency and widespread application perspective of the most large relational databases. The discovery algorithm and an example are discussed in details.