Record subtyping in flexible relations by means of attribute dependencies

Abstract

The model of flexible relations supports heterogeneous sets of tuples in a strongly typed way. The elegance of the standard relational model is presented by using a single, generic scheme constructor. In each model supporting structural variants the shape of some part of a heterogeneous scheme may be determined by the contents of some other part of the scheme. We formalize this relationship by a certain kind of integrity constraint we have called "attribute dependency" (AD). We motivate how ADs can be used, besides their application in type and integrity checking, to incorporate record subtyping into our extended relational model. Moreover, we show that ADs yield a stronger assertion than the traditional record subtyping rub as they consider interdependencies among refinements. We discuss how ADs are related to query processing and how they may help to identify redundant operations.<>