A rough set model based on Formal Concept Analysis in complex information systems

Abstract

As a relatively new theory, Formal Concept Analysis, also called concept lattice, is a kind of mathematical tool for analyzing and processing binary relation in essence, initiated by German scholar Wille in 1982. At present, the theory has been studied extensively and found wide applications in fields like machine learning, software engineering, information retrieval, etc. Normally, in some complex information systems, the binary relation of domain of any attribute is a more general binary relation, which does not satisfies common properties such as reflexivity, transitivity or symmetry. Meanwhile, because the large differences may exist in different attributes, there may exist completely different types of binary relations in different domains of attributes. As is stated above, by introducing concept lattice into rough set theory, this paper expands equivalence relation, dominance relation, similarity relation etc. to a more general binary relation, and then discusses the granularity model for the general binary relation mentioned above. Based on this, a new knowledge acquisition model based on concept lattice is proposed. In the paper, a algebraic structure can be drawn from a complex information system, which is a lattice in essence. In additional, the paper mainly probes into attribute reduct, core in complicated information systems; Finally, how to eliminate redundant rules in decision tables is studied.