Attribute Reduction In Fuzzy Formal Contexts And Knowledge Acquisition Of Multi-source Formal Contexts

Concept lattice theory is an effective tool for knowledge representation and knowledge discovery.It has been widely used in decision analysis,machine learning,data mining and other fields.Attribute reduction of formal contexts is an important topic in formal concept analysis.Due to the complexity of the structure of fuzzy concept lattices,attribute reduction of fuzzy formal contexts has become a difficult problem when studying fuzzy concept lattice theory.Besides,knowledge discovery of multi-source data is an important problem to be solved in the field of big data analysis.The studies about knowledge discovery in multi-source formal context are relative few.On the basis of related research,this paper is devoted to the discussion of attribute reduction method in fuzzy formal context and knowledge acquisition approach in multi-source formal contexts.The main contents of this paper include:1.The notions of consistent set and attribute reduction in fuzzy formal context are proposed in this paper.Some judgement theorems for a set of attributes to be consistent are presented.By using the discernibility function composed of discernibility attributes between related L-fuzzy formal concepts,an approach to compute attribute reductions is proposed.2.A method to combine multi-source formal contexts into an L-fuzzy formal context is presented.The structures of concepts in fusion L-formal context are described.The relationship between variable threshold concepts in fusion L-formal context and formal concepts in single source formal context is discussed.The fusion of multi-source fuzzy formal context is discussed by means of Cartesian product of complete residuated lattices.It is proved that the fuzzy concepts in the fusion fuzzy formal context of multi-source fuzzy formal context and the fuzzy concepts in the single source fuzzy formal context can be mutually determined.3.The granular reduction method of the fusion L-formal context based on the crisp-crisp variable threshold concept is presented.The definition of granular reduction of fusion L-formal context is proposed,and some judgement theorems of granular consistent set are presented.The relationship between the granular reduction of this fuzzy formal context and that of single source formal context are investigated.