Abstract
We propose a new approach to the design of relational database schemes. The main features of the approach are: (a) A combination of the traditional decomposition and synthesis approaches, thus allowing the use of both functional and multivalued dependencies. (b) Separation of structural dependencies relevant for the design process from integrity constraints, i. e. , constraints that do not bear any structural information about the data and must therefore be discarded at the design stage. This separation is supported by a simple syntactic test filtering out nonstructural dependencies. (c) Automatic correction of schemes that lack certain desirable properties.
Original language | English |
---|---|
Pages | 196-207 |
Number of pages | 12 |
State | Published - 1984 |