Efficient processing of interactive relational data base queries expressed in logic

Abstract

Relational database retrieval is viewed as a special case of deduction in logic. It is argued that expressing a query in logic clarifies the problems involved in processing it efficiently (query optimisationn). The paper describes a simple but effective strategy for planning a query so that it can be efficiently executed by the elementary deductive mechanism provided in the programming language Prolog. This planning algorithm has been implemented as part of a natural language question answering system, called Chat-80. The Chat-80 method of query planning and execution is compared with the strategies used in other relational database systems, particularly Ingres and System R.