A knowledge management framework that supports evolution of configurable products

RHOMBOS Bibliografische Information der Deutschen Nationalbibliothek Die Deutsche Nationalbibliothek verzeichnet diese Publikation in der Deutschen Nationalbibliografie; detaillierte biblio-grafische Daten sind im Internet über http:// About 25 years ago AI researchers developed the first framework for structure-based configuration, i.e. for building complex technical systems from computer-based models of components and of their possible combinations. Since then, configuration technology has entered many application areas ranging from automobile to software production, and computer-based models are applied throughout a products life cycle from design and documentation to configuration and maintenance. With todays pressure on rapid innovations and growing model repertoires, the task of maintaining complex model databases has become a challenge. How do changes of component properties affect existing or planned products? What kind of tool can support or maybe even guarantee orderly maintenance and evolution of a model database? The doctoral thesis of Thorsten Krebs provides answers to these questions. Building on a long tradition of research into configuration systems and formal knowledge representation in the Cognitive Systems Group at the University of Hamburg, the author develops a well-defined set of change operations, each furnished with preconditions and postconditions to guarantee consistency of the model database. It is interesting to compare this work to current research into ontology maintenance based on the Semantic Web ontology language OWL-DL. In both cases, description logics (DLs) provide the formal basis for knowledge representation and consistency preservation is a recognized issue. The work on product evolution , however, is in several ways more ambitious than ontology maintenance on OWL: Product evolution deals with processes which can handle changes and simultaneously maintain consistency, while typical tools for ontology maintenance (DL reasoners) can only check a static ontology for consistency. It is also interesting to observe that configuration technology requires constraints as an essential part of its knowledge representation, while OWL and DL reasoners are very restricted regarding the use of constraints. The expressivity of standardized ontology languages has to be significantly extended before they can handle product evolution. While striving to obtain formally founded solutions, the author is in the end committed to a user-oriented approach where practical considerations such vi as an unconstrained order of operations and manageable complexity are important. The approach presented in this thesis can therefore be considered as a first instantiation of a product evolution framework with a practical significance similar to the one of configuration systems 25 years ago. Foreword So, that's …