论文部分内容阅读
在知识发现过程中用户感兴趣的往往是一些高层次、适当概括的简化信息,面向属性的归纳是目前主要的数据归约方法,一般是仅考虑原始数据所提供简单的统计信息.本文提出的基于量化扩展概念格的属性归纳算法,采用概念的爬升进行相应的泛化来完成多层、多属性归纳.与面向属性归纳算法比较,该算法的泛化路径不是唯一的,在量化扩展概念格的哈斯图中容易找到合适的泛化路径和阈值,得到满足用户要求合理的属性归纳结果,以提供用户所需的不同粒度的知识.
In the process of knowledge discovery, the users are often interested in some simple, high-level and generalized simplified information. Attribute-oriented induction is the main method of data reduction at present, and generally only considers the simple statistical information provided by the original data. Based on the attribute extension algorithm of quantitative extension concept lattice, the generalization of concept climb is used to complete the multi-level and multi-attribute induction.Compared with the attribute-oriented induction algorithm, the generalization path of the algorithm is not unique, Hastu easy to find the appropriate generalization path and threshold, to meet the user requirements reasonable attribute induction results, in order to provide users with different granularity of knowledge.