This is one of my all-time favorite database papers---and I'd recommend it highly to anyone working on object extensions to relational database systems. In a nutshell, this paper extended the relational model and a relational query language (Quel) to provide for more strongly typed tables, generalization, references, path expressions, and set-valued attributes. What makes this paper such a gem (pun intended!) is that its extensions are remarkably clean, simple, and natural. While some ``relational bigots'' will argue even today that relations and objects are like oil and water, my view is that Zaniolo in many ways proved them wrong through his GEM research almost 15 years ago. Some of the highlights of this paper are its introduction of the dot-notation for path expressions, its clear treatment of the interplay between null values and path expressions, its unique type system, and its treatment of sets. This paper strongly influenced the object-oriented data model and query language work that we did in the context of the EXODUS project at Wisconsin; it continues to influence my view of the world in my current work on object-relational extensions for DB2 UDB and SQL3.
a service of Schloss Dagstuhl - Leibniz Center for Informatics