Semantic optimisation in datalog programs

Abstract

Datalog is the fusion of Prolog and Database technologies aimed at producing an efficient, logic-based, declarative language for databases. This fusion takes the best of logic programming for the syntax of Datalog, and the best of database systems for the operational part of Datalog. As is the case with all declarative languages, optimisation is necessary to improve the efficiency of programs. Semantic optimisation uses meta-knowledge describing the data in the database to optimise queries and rules, aiming to reduce the resources required to answer queries. In this thesis, I analyse prior work that has been done on semantic optimisation and then propose an optimisation system for Datalog that includes optimisation of recursive programs and a semantic knowledge management module. A language, DatalogiC, which is an extension of Datalog that allows semantic knowledge to be expressed, has also been devised as an implementation vehicle. Finally, empirical results concerning the benefits of semantic optimisation are reported.