论文部分内容阅读
提出一种基于属性分辨度的不完备决策表规则提取算法,它是一种例化方向的方法.首先从空集开始,逐步选择当前最重要的条件属性对对象集分类,从广义决策值唯一的相容块提取确定规则,从其他的相容块提取不确定规则;然后设计属性必要性判断步骤去除每条规则的冗余属性;最后通过规则约简过程来简化所获得的规则,增强规则的泛化能力.实验结果表明,所提出的算法效率更高,并且所获得的规则简洁有效.
This paper presents an algorithm of incomplete decision table extraction based on attribute resolution, which is a method to instantiate the direction.First, from the empty set, the most important condition attributes are selected step by step to classify the object set, Then extract the uncertain rules from other compatible blocks; then design the attribute necessity judgment step to remove the redundant attributes of each rule; finally, simplify the rules obtained by the rule reduction process and enhance the rules The experimental results show that the proposed algorithm is more efficient and the rules obtained are concise and effective.