基于概念格的关联规则发现

CONCEPT LATTICE BASED ASSOCIATION RULE DISCOVERY

关联规则是数据库中有价值的一类规律 ,国内外学者已经对它进行了广泛的研究 .由二元关系导出的概念格作为一种非常有用的形式化工具 ,体现了概念内涵和外延的统一 ,反映了对象和特征间的联系以及概念的泛化与例化关系 ,因此非常适于发现数据中潜在的概念 .本文分析了概念格的关联规则提取之间的关系 ,根据需要对格结构进行了相应的修改 ,提出了概念格结点内涵缩减概念 ,给出了相应的渐进式生成算法和基于概念格的关联规则提取算法 。

Association rule is a class of valuable regularity in databases, which have been widely studied. Concept lattice, induced by a binary relation, is a very useful formal tool. It realizes the unification of concept intension and concept extension, represents the association between objects and features, and reflects the generalization and the specialization among concepts, so is fit for discovering the potential concept below the data. This paper first analyzes the relationship between concept lattice and association rule discovery, then modifies the structure of node according to the requirement, gives out the formal definition of intension reduct presents algorithms for constructing the corresponding lattice incrementally and extracting association rule, and uses some theorems and properties to reduce the number of discovered rules.