Abstract: The current rule extraction algorithms are summarized. A two-stage rule extraction algorithm is proposed. This algorithm has covering and confidence factors to eliminate noise rules produced by noise data. Supporting degree is presented to measure the algorithm validity. Through running an example, the algorithm is shown to be able to eliminate noise rules and make the rules extracted simpler and more practical

Key words: Rough sets; Rule extraction; Noise rules; Supporting degree

摘要： 在总结分析现有规则提取算法的基础上，提出了一种新的带覆盖度和置信度因子的两阶段规则提取算法，以解决噪音数据产生的噪音规则问题，并提出了一个衡量算法有效性的指标——支持度。实例表明，该算法能够有效地去除噪音规则，使提取出的规则更加精练实用。

关键词: 粗糙集；规则提取；噪音规则；支持度