Abstract:

In spite of its power in terms of data manipulation and integrity support, the relational model has been considered as not having too much to offer for modelling semantics of the real world applications. Consequently, a number of semantic data models have been proposed to design relational databases. Even more radical approaches recommend giving up entirely the normalization achievements and urge using only OO models and tools. This paper proposes a graphical-oriented methodology for database design based on Bernsteins’s theory of normalization by synthesis, on functional and inclusion dependencies, on Smith’s and also Ausiello, D’Atri, and Sacca’ formalisms. Articulated entirely within the relational model, the methodology, lacking a sound theoretical foundation - this lack is deemed to attract practitioners -, has been tested for the last five years in some small and medium sized applications, proving to be useful and understandable.