On Concept Lattices and Implication Bases from Reduced Contexts

Our paper introduces well-known methods for compressing formal context and focuses on concept lattices and attribute implication base changes of compressed formal contexts. In this paper Singular Value Decomposition and Non-negative Matrix Factorisation methods for compressing formal context are discussed. Computing concept lattices from reduced formal contexts results in a smaller number of concepts (with respect to the original lattice). Similarly, we present results of experiments in which we show a way to control smoothly the size of generated Guigues-Duquenne bases and provide some noise resistance for the basis construction process.