面向属性概念格基于覆盖的压缩
Covering-Based Reduction of Property-Oriented Concept Lattices
-
摘要: 提出了一种直接从形式背景出发,获得简化面向属性概念格的方法。针对一个形式背景,引入对象集和属性集上的相似度,产生对象邻域与属性邻域,进而根据对象的相似程度或属性的相似程度来调整对象邻域或属性邻域的大小,控制面向属性概念格节点的个数,实现对面向属性概念格的动态压缩。该方法可以分别从对象的覆盖和属性的覆盖两个角度压缩面向属性概念格中节点的个数,且压缩后的概念集为原概念集的子集,从而达到简化知识库的效果。Abstract: A method to reduce the property-oriented concept lattice of a formal context directly is proposed in this paper. The similarity degree between object sets and the similarity degree between attribute sets are firstly introduced, and then, object neighborhood and attribute neighborhood are created accordingly. The sizes of object neighborhood or the attribute neighborhood are adjusted by the similarity degrees, thus, the number of the property-oriented concepts can be controlled and the property-oriented concept lattice is compressed. Using this method, we can compress the property-oriented concept lattice from the viewpoints of object covering and attribute covering, and the most important is that the reduced lattice is a subset of the original one.