A Variant of Earley Deduction With Partial Evaluation

TL;DR

This paper introduces a novel algorithm for query evaluation in function-free Datalog programs, leveraging Earley Deduction and partial evaluation to generate finite automata, improving efficiency over existing methods in certain cases.

Contribution

It presents a new variant of Earley Deduction that incorporates partial evaluation to produce finite automata for query processing, enhancing efficiency.

Findings

The new method can process multiple deduction steps simultaneously.

It outperforms SLDMagic and Magic Set methods in specific scenarios.

Finite automata are effectively generated for query evaluation.

Abstract

We present an algorithm for query evaluation given a logic program consisting of function-free Datalog rules. It is based on Earley Deduction [4, 6] and uses a partial evaluation similar to the one we devel oped for our SLDMagic method [1]. With this, finite automata modeling the evaluation of given queries are generated. In certain cases, the new method is more efficient than SLDMagic and the standard Magic Set method since it can process several deduction steps as one.