Abstract: A decision rule mining algorithm based on rough information vector is proposed in inconsistent information system.The brief decision rules according with the demands of threshold is directly mined from decision tables based on rough information vector.In this process，decision support ability of condition vector to decision vector is used，and the concrete decision support ability of condition attribute value is not losing.The consistent rules set and default rules set are mined in the case of the each simplified level of condition attribute.The feasibility of algorithm is indicated using theoretical analysis and the concrete instance.

Key words: inconsistent information system, decision rules, rough information vector

摘要： 针对不一致信息系统中决策规则获取问题，提出了一种基于粗糙信息向量方法的决策规则挖掘算法。基于粗糙信息向量，利用条件向量对决策向量的决策支持能力，直接从决策表中挖掘出符合阈值要求的尽可能简洁的决策规则，且不损失条件属性值的决策支持能力。利用该算法可以挖掘出决策系统中条件属性在各个简化层次情况下的确定性规则和缺省规则集合。理论分析和实例表明该算法在不一致信息系统中的决策规则获取上是可行的。

关键词: 不一致信息系统, 决策规则, 粗糙信息向量