不完备信息系统中基于相似关系的知识约简

Knowledge Reductions in Incomplete Information Systems Based on Similarity Relation

以具有丢失型未知属性值的不完备信息系统为研究对象,根据非对称相似关系,讨论了知识约简问题。在不完备决策系统中,引入了近似、粗糙分布约简以及广义决策约简,讨论了它们之间的相互关系,给出了近似分布约简的判定定理、可辨识矩阵以及约简公式,并进行了实例分析,为从不完备信息系统中获取知识提供了新的理论基础与操作手段。

In this paper, the incomplete information system in which all unknown values are looked as lost is deeply investigated. In such incomplete information systems, approach to knowledge reduction based on similarity relation is studied. Moreover, the approximate and rough distribution reducts, generalized decision reduct are introduced into in- complete decision system, the judgment theorems, discernibility matrixes and reduction formulas with respect to ap- proximate distribution reducts are obtained. Finally, an illustrative example is analyzed. These results are meaningful both in the theory and in applications for rules＇ acquisition in incomplete decision systems.