Local logical disjunction double-quantitative rough sets

Abstract Local rough sets as a generalization of classical rough sets not only inherit the advantages of classical rough sets which can handle imprecise, fuzzy and uncertain data, but also break through the limitation of classical rough sets requiring large amount of labeled data. The existing researches on local rough sets mainly use the relative quantitative information between a target concept and equivalence classes of those objects contained in the target concept to approximate the target concept. This ignores the information differences of equivalence classes concerned containing the relevant concept, namely the absolute quantitative information. We propose Local Logical Disjunction Double-quantitative Rough Sets (LLDDRS) model based on the importance, completeness and complementary nature of the relative and absolute quantitative information to describe an approximation space. This provides an effective tool for discovering knowledge and making decisions in relation to large data sets. In this paper we first study the important properties, optimal computing of rough regions and decision rules of the LLDDRS model. Then we explore the relationships of the proposed LLDDRS model and other representative models. Finally, we present experimental comparisons showing the computational efficiency and approximate accuracy of the LLDDRS model in concept approximation.