Algorithmen zur Begriffsanalyse und ihre Anwendung bei Softwarebibliotheken
by
Christian Lindig
Technische Universität Braunschweig, November 1999.
Abstract
Formal concept analysis is an algebraic theory concerning binary relations and closely related complete lattices of so-called concepts. This thesis presents algorithms and data structures to compute concepts and their lattice structure. Since, in the worst case, the number of concepts can grow exponentially with the size of a relation, the complexity of algorithms was given special attention. The speed of algorithms as well as the actual size of concept lattices was tested with a large number of test cases. They show that for practical applications the performance of algorithms and the number of concepts depends only quadratically on the size of the relation. As an application for concept analysis the organization of a library of re-usable software components is proposed. Component based software reuse aims to raise software quality and development productivity by re-using already successfully developed components. The proposed method combines good maintainability of the component library with strong navigation support for the user. The search tool uses the concept lattice as a data structure that is computed once for the library. The lattice essentially contains all decisions a user can take while searching for a component and thus supports efficient searching. Additionally the lattice permits reasoning about the quality of the indexing method used on the components.
BibTeX Entry
@phdthesis{lindig-phd-1999,
title = "Algorithmen zur Begriffsanalyse und ihre Anwendung bei Softwarebibliotheken",
author = "Christian Lindig",
year = "1999",
month = nov,
location = "D--38106 Braunschweig, Germany",
school = "Technische Universität Braunschweig",
}