Interpreting a reconstructed relational calculus (extended abstract)

Abstract

This paper describes a method for answering all relational calculus queries under the assumption that the domain of data values is sufficiently large. The method extends recent theoretical results that use extended relation representations to answer domain dependent queries, without the use of auxilliary variables or invented constants or an explicit enumeration of the active domain. The method is shown to be logically correct and to have polynomial data complexity. By identifying relational algebra operations with relational calculus queries this approach extends relational algebra to a full boolean algebra, where intersection, union, and difference are defined between any two relations, whether or not they are union compatible. An example illustrates that this approach can be useful in distributed query optimization.