论文部分内容阅读
基于差别矩阵属性约简方法获得的约简并不能保证与正区域约简一致,并且在处理高维大数据集时将消耗过多的存储空间.为此,首先对决策表进行简化,引入属性布尔差别矩阵及其核属性和属性约简定义,同时证明了该约简与正区域约简是一致的.然后,基于属性布尔差别矩阵,设计正向启发式属性约简算法;为了进一步减少算法的空间开销,引入Swapping技术,并给出反向启发式属性约简算法.最后,实例和实验结果表明所提出的约简算法是正确的、高效的.
The reduction based on the discernibility matrix attribute reduction method does not guarantee consistency with the positive region reduction and consumes too much storage space when dealing with high-dimensional large data sets.To this end, we first simplify the decision table and introduce attributes Boolean difference matrices and their definitions of core attributes and attribute reductions, at the same time, it is proved that the reductions are consistent with the reductions in the positive region. Then, based on the Boolean difference matrix, a heuristic attribute reduction algorithm is designed. In order to further reduce the algorithm , The Swapping technique is introduced, and the inverse heuristic attribute reduction algorithm is given.Finally, the examples and experimental results show that the proposed reduction algorithm is correct and efficient.