论文部分内容阅读
该文针对RoughSet理论中属性约简和值约简这两个重要问题进行了研究,提出了一种借助于可辨识矩阵(discernibilitymatrix)和数学逻辑运算得到最佳属性约简的新方法.同时,借助该矩阵还可以方便地构造基于RoushSet理论的多变量决策树.另外,对目前广泛采用的一种值约简策略进行了改进,最终使得到的规则进一步简化.
In this paper, two important problems of attribute reduction and value reduction in RoughSet theory are studied. A new method of optimal attribute reduction based on discernibility matrix and mathematical logic operation is proposed. At the same time, with the help of this matrix, it is also easy to construct a multivariate decision tree based on RoushSet theory. In addition, a value reduction strategy that is widely used today is improved, and finally the rules obtained are further simplified.