On efficient factorization of standard fuzzy concept lattices and attribute-oriented fuzzy concept lattices

We propose two ways to make factorization of attribute-oriented fuzzy concept lattices recently studied by Ciobanu & Văideanu more efficient. Firstly, we show that the blocks of the corresponding factor lattice are determined by a particular fuzzy interior operator. This allows us to compute an attribute-oriented fuzzy concept lattice by intents. Such a computation requires fewer traversals through data, as the number of the attributes is usually smaller than the number of objects. Secondly, we scale the input data in such a way that the factor lattice can be computed as a one-sided fuzzy-crisp concept lattice. Our experiments indicate that both methods lead to a significant improvement in efficiency of computation of the factor lattice. In addition, we improve and extend the theory provided by Ciobanu & Văideanu.