A CORE PROTOTYPE IMPLEMENTATION OF AN EXTENDED RELATIONAL DATABASE LANGUAGE FOR EFFICIENT MANIPULATION OF CONTAINMENT AND COMPOSITE STRUCTURES

Abstract

The core prototype implementation of an extended relational data base system is described. The full data model allows for restricted unnormalized relations. Essentially attributes that are sets or lists of atomic attributes are allowed but not attributes that are relations. There is support for both inheritance and composite entities. Manipulation languages are SQL and a separate new SQL language subset for efficient manipulation of containment structures. The language subset has a predicate calculus expression structure, but allows the use of genitive relations to model containment relationships and facilitate application of natural quantifiers to containment sets. A genitive relation is a relation that is the equivalent to the genitive case grammatical construct in natural language. In the query processor a query decomposition technique was employed. Declarative language expressions involving naturally quantified genitive relation clauses for one-to-many relationships, both composite and non composite, both recursive and nonrecursive, and for subset genitive relations, can be reduced. The natural quantifiers are the quantifiers of natural language and the prototype implements 14 basic types including universal and existential quantifiers. The query processor translates declarative language expressions to Extended Relational Algebra (ERA) routines. ERA differs from conventional relational algebra in that it has a possibility join operation, as well as two operations involving quantifiers, namely group select and subgroup select operations.