A New Method for Fuzzy Formal Concept Analysis
Siyao Zheng, Yiming Zhou, Trevor Martin · 2009
In this paper, a fuzzy scaling method to multi-value attributes, the corresponding model of fuzzy concepts and fuzzy concept lattice are proposed. Traditional formal concept analysis (FCA) deals with only binary context, and provides only crisp scaling method to deal with multi-value context, which is not an appropriate granulation for human reasoning. This paper first proposes a method of scaling multi-value attributes into fuzzy attributes using the concept of linguistic variables, then defines fuzzy concept, which has a fuzzy set of objects for extension and a crisp set of scale attributes for intension, and fuzzy concept lattice, which is a partial order set of fuzzy concepts. An efficient method of extracting fuzzy concepts from multi-value context in a crisp style is also proposed.