%0 Journal Article %T 基于决策分类的分块差别矩阵及其求核算法<br>Block discernibility matrix based on decision classification and its algorithm finding the core %A 左芝翠 %A 张贤勇 %A 莫智文 %A 冯林< %A br> %A ZUO Zhi-cui %A ZHANG Xian-yong %A MO Zhi-wen %A FENG Lin %J 山东大学学报（理学版） %D 2018 %R 10.6040/j.issn.1671-9352.4.2018.121 %X 摘要： 属性约简是粗糙集理论进行数据挖掘的基本途径, 相关算法主要基于核。 核的差别矩阵表示及相关求核计算具有重要意义, 但已有的差别矩阵及其求核算法还具有时空局限性。对此, 依据差别矩阵的稀疏性与大规模性, 提出基于决策分类的分块差别矩阵及其求核算法, 直接地将决策分类信息融入形式结构与问题求解。 首先, 基于决策分类来定义分块差别矩阵, 设计其计算算法; 其次, 基于分块差别矩阵, 确定核的内涵与算法; 最后, 进行实例分析与实验验证, 说明所建方法的有效性。基于决策分类的分块差别矩阵有效地实施了信息提取与维度降低, 相关的求核算法较好地减少了差别矩阵求核算法的时空复杂性。<br>Abstract: Attribute reduction is the fundamental approach of rough set theory to implement data mining, and its relevant algorithms are mainly based on the core. For the core, both its representation of the discernibility matrix and its calculation for finding the core exhibit important significance, but the existing discernibility matrix and its core algorithm have time and space limitations. According to the sparsity and large scale of the discernibility matrix, the block discernibility matrix based on the decision classification and its algorithm finding the core are proposed, and thus the decision classification information is directly applied to the form structure and problem solving. At first, the block discernibility matrix is defined by the decision classification, and its calculation algorithm is achieved. Then, based on the block discernibility matrix, the essence and algorithm of the core are provided. Finally, the proposed methods effectiveness is verified by the example and experiment. The block discernibility matrix based on the decision classification effectively implements the information extraction and dimensionality reduction, so its relevant algorithm finding the core well decreases the time and space complexities of the corresponding algorithm based on the discernibility matrix %K 粗糙集 %K 核 %K 差别矩阵 %K 属性约简 %K 决策分类 %K 分块差别矩阵 %K < %K br> %K attribute reduction %K core %K block discernibility matrix %K decision classification %K discernibility matrix %K rough set %U http://lxbwk.njournal.sdu.edu.cn/CN/10.6040/j.issn.1671-9352.4.2018.121