中国科学院重庆绿色智能技术研究院机构知识库

针对面向领域用户的决策规则挖掘问题，用属性序描述领域用户的需求和兴趣，模拟人脑分辨事物的过程，提出了一种属性序下的分层递阶决策规则挖掘算法。该算法在给定属性序下输出的决策规则集不仅具有唯一性，且对任意待识别样本不会作出矛盾的决策。实例和仿真实验结果表明了算法的有效性和可行性。

For a decision rules mining problem of user's requirements, the attribute order is used as the description of user's requirements, and a hierarchical algorithm of decision rules mining based on the attribute order is proposed. For a given attribute order, the set of decision rules not only is unique, but also the inconsistent decision can not be done for any new object. Simulation experimental results show the effectiveness and feasibility of the algorithm.