粗糙集分类算法中的近似决策规则和规则匹配方法

粗糙集分类算法在应用标准决策规则进行新对象分类时,经常碰到决策规则与新对象不完全匹配的情况。因此,近似决策规则和部分匹配方法常用于提高决策规则与新对象匹配的可能性。本文在概述和比较两种近似决策规则生成算法的基础上,以一个文本分类系统为例,提出了一种综合的、更有效的近似决策规则生成算法。文章还介绍了几种通用的规则匹配方法,提出了一系列实用的完全匹配和部分匹配公式。实验表明,新提出的近似决策规则生成算法和规则匹配公式能够有效地提高决策规则与新对象的匹配可能性与准确性。

Approximate Decision Rules and Matching Rules in Rough Set Based Classification Algorithms

In most cases,the decision rules inducted by rough set models are unacceptable as laws to classify new ob- jects. Approkimate decision rules and partial matching rules are proposed to overcome this problem. This paper dis- cusses two typical algorithms for the generation of approximate rules and comparatively analyzes their performance as proven by one case study. Furthermore, one more efficient algorithm is developed based on the two algorithms. This paper also describes the general measures used for matching rules,and a set of formulae are defined for complete matching and partial matching of decision rules according to dependency coefficient in rough set theory. The experi- ments show that the proposed approximation algorithm and measures for matching rules can further improve the matching possibility and correctness of basic decision rules generated based on rough set theory.