Restrictions on Concept Lattices for Pattern Management

This paper addresses the problem of pattern management in the framework of formal concept analysis using restrictions on objects or attributes of a given data set. Patterns are pieces of information/knowledge with a concise description that can be obtained from data using data mining techniques. These can be clusters or bi-clusters, implications or association rules, and so on. Even for relatively small data collections, the set of discovered patterns could be very large and therefore difficult to handle. In this paper we propose efficient methods to conduct the projection of a concept lattice on a set of attributes. The selection of a concept lattice on a set of objects is done dually.