论文部分内容阅读
许多研究表明概念格是数据分析与规则提取的一种有效工具．本文首先提出一种在对象与描述符数目较多、概念聚类具有一定规模条件下，在已建造好的概念格上有效地提取规则的算法．这种方法主要依据格结点的直接泛化来产生相应无冗余规则；然后改进了一种渐进式更新概念格与相应Ｈａｓｓｅ图的算法，并将之应用于渐进式提取规则．目前，这些方法已用于我们所开发的数据库知识发现工具原型系统中．
Many studies show that concept lattice is an effective tool for data analysis and rule extraction. In this paper, we first propose an algorithm for efficiently extracting rules from constructed concept lattices with a large number of objects and descriptors and a certain scale of concept clustering. This method mainly produces the corresponding non-redundant rules according to the direct generalization of the lattice nodes. Then an improved algorithm of updating the concept lattice and the corresponding Hasse graph is improved and applied to the gradual extraction rules. At present, these methods have been used in the database knowledge discovery tool prototype system we developed.