(An) approach to efficient database design incorporation usage information

Abstract

In this thesis, a new approach using usage information is presented for designing an efficient database schema. For conceptual design, the concept of qualification dependencies (QDs) and a rule based algorithm to derive interactively a conceptual database model, i.e., an ER model, based on the QDs are presented. The QDs can be identified based only on the natural relationships between attributes perceived by the designer, thereby allowing to design an ER model which is fairly compatible to the designer``s understanding of the real world. The architecture of an intelligent database design system (IDDS) which implements the ER design rules is also presented. For logical design, a new usage dependency model (UDM) is presented which tries to benefit from both data dependencies and usage information. The model extends the traditional data dependency model of relational tehory by combining the natural (functional) and usage views of data. The combined application of usage depends which is a process-oriented view of data correspondence, and data dependency, which is a nonprocessing or natural view of data dependency, which is a nonprocessing or natural view of data correspondence, can produce a normalized database design that is efficient for current processing requirements as well as one which is adaptable to future requirements. A comprehensive algorithm designing an efficient database schema based on the UDM is also presneted, which can be easily incorporated in the existing database design methodologies. It is also shown that the concept of Qds allows to integrate the conceptual and logical design in designing an efficient database schema in that an efficient relational schema can be directly derived from the Er model which is constructed based on the concept of QDs.