Stored and Inherited Relations

TL;DR

This paper revisits a 1992 proposal for a third relational construct combining stored and inherited attributes, demonstrating its potential to simplify database schemas and improve fidelity to real-world data with minimal overhead.

Contribution

The paper reintroduces and advocates for a third relational construct that merges stored and inherited attributes, enhancing database schema expressiveness and practicality.

Findings

Using the construct can reduce query complexity.

It may eliminate the need for auxiliary views.

Implementation incurs minimal storage and processing overhead.

Abstract

The universally applied Codd's relational model has two constructs: a stored relation, with stored attributes only and a view, only with the inherited ones. In 1992, we have proposed third construct, mixing both types of attributes. Examples showed the idea attractive. No one followed however. We now revisit our proposal. We show that a relational database scheme using also our construct may be more faithful to reality. It may spare the logical navigation or complex value expressions to queries. It may also avoid auxiliary views, often necessary in practice at present. Better late than never, existing DBSs should easily accommodate our proposal, with almost no storage and processing overhead.