| The aim of this dissertation is to propose new fast and efficient algorithms for formal concept analysis. Formal concept analysis is a method of knowledge extraction from incidence object-attribute data which has its foundations in theory of ordered sets. The basic output of formal concept analysis is a set of formal concepts, i.e., particular conceptual clusters present in the input data. From the computational point of view, formal concept analysis is very demanding and, until recently, it lacked fast algorithms, and it was therefore applicable only for small or medium sized datasets. In order to develop fast algorithms for formal concept analysis that can be used to process large datasets, we focus on all aspects of the algorithm design. Especially, we focus on the design of efficient parallel and distributed algorithms. The family of new algorithms introduced in this Thesis outperforms by orders of magnitude other state-of-the-art algorithms for formal concept analysis. |
