摘要： 粗糙集属性分区数的变化会影响属性重要性和属性对决策属性的支持度。该文对知识表示系统的数据相关性进行分析，综合考虑系统的泛化能力，提出能生成确定性控制规则的决策模型，给出决策模型中属性分区数求取以及属性相对约减产生的判据与算法实现。实验结果表明，该算法简洁有效，验证了决策模型的准确性与实用性。

关键词: 决策模型, 确定性控制规则, 属性分区, 属性重要性, 支持度

Abstract: Different attributes partitions affect about attribute significance and attributes support degree to decision attribute. This paper analyzes the relationship among data in knowledge represent system based on rough set. Generalization of system is combined to consider. A decision model is presented for generating certain control rules. Judgment and algorithm are introduced to compute attributes partitions in decision model and find relative attributes reduction in decision model. Experimental results show conciseness and efficiency of algorithms and preciseness and utility of decision model.

Key words: decision model, certain control rules, attribute partition, attribute significance, support degree

中图分类号: 

• TP18