Formal concept analysis via multi-granulation attributes

Abstract

The study of concept lattices, property oriented concept lattices and object oriented concept lattices provides complementary conceptual structures, which can be used to search, analyze and extract information from large data sets. In the view of multi-granular computing, this paper proposes a novel approach for formal concept analysis via multi-granulation attributes. Based on a formal context and an equivalence relation on its attribute set, we construct a multi-granulation formal context. The related attribute oriented concept lattices are investigated. Some important properties are examined. We also provide methods to generate attribute oriented concept lattices associated with a multi-granulation formal context. This study can be used to analyze concepts and sub-concepts on attributes, to mine multi-level association rules and cross-level association rules originated from practice application, such as in our real-life market basket analysis.