Computing graph-based lattices from smallest projections

Abstract

From the mathematical perspective, lattices of closed descriptions, which arise often in practical applications can be reduced to concept lattices by means of the Basic Theorem of Formal Concept Analysis (FCA). From the computational perspective, in many cases it is more advantageous to process closed descriptions and their lattices directly, without reducing them to concept lattices. Here a method for computing lattices with descriptions given by sets of graphs, starting with rough approximations is considered and compared to previous approaches. © 2011 Springer-Verlag.