Data Mining Via Lattice Machine

Shi WANG,Hui WANG,Wen GAO

DOI: https://doi.org/10.3321/j.issn:0254-4164.2000.06.002

2000-01-01

Jisuanji Xuebao/Chinese Journal of Computers

Abstract:This paper introduces a novel approach to data mining based on Lattice Machine. This method is a way from data reduction to data mining by lattice machine. Hyper relations are a generalization of conventional database relations in the sense that sets of values are allowed as tuple entries. The collection of hyper relations can be naturally transformed into a complete Boolean algebra, and so for any collection of hyper tuples its unique least upper bound (lub) can be found as a reduction of it. The process is to find a set of elements i.e. interior cover in the lattice, which is closest to the least upper bound of the labeled elements, and which is such that all elements in the interior cover are ″equilabeled″. This ″equilabeledness″ guarantees that the originally unlabeled elements. Authors tend to find the interior cover, the subset of internal elements covered by the lub, after finding the interior cover, which can in turn be used as an approach to data mining. This approach can be easily used to represent rules, and naturally avoid the missing value.