AFS-Based Formal Concept Analysis on Multi-valued Context

Non-symmetric indiscernibility preorders, arising naturally from hierarchically structured data, are general methods of conceptual scaling for multi-valued data. Based on the original idea of Ganter on non-symmetric indiscernibility preorders and the AFS (Axiomatic Fuzzy Sets) theory, this paper investigates a multi-valued formal concept analysis, in which multi-granule concept lattice is established on multi-valued context, and the membership function is directly determined from the original data of the multi-valued context. Compared with symmetric indiscernibility relations, the main advantage of the proposed method is capable of dealing with information table with fuzzy attributes, Boolean attributes, and intuition order attributes, and describing the uncertainty relations between the objects and the attributes for formal context with complex data.