NP Datalog: a Logic Language for Expressing NP Search and Optimization Problems

TL;DR

This paper introduces NP Datalog, a logic language extending Datalog with constructs for expressing NP search and optimization problems, and demonstrates its effectiveness through a prototype system and experiments.

Contribution

It presents NP Datalog, a novel logic language for expressing NP problems, and a system prototype translating queries into efficient OPL programs.

Findings

NP Datalog captures the power of Datalog with unstratified negation.

The system translates NP Datalog queries into OPL programs.

Experiments show the approach is effective and efficient.

Abstract

This paper presents a logic language for expressing NP search and optimization problems. Specifically, first a language obtained by extending (positive) Datalog with intuitive and efficient constructs (namely, stratified negation, constraints and exclusive disjunction) is introduced. Next, a further restricted language only using a restricted form of disjunction to define (non-deterministically) subsets (or partitions) of relations is investigated. This language, called NP Datalog, captures the power of Datalog with unstratified negation in expressing search and optimization problems. A system prototype implementing NP Datalog is presented. The system translates NP Datalog queries into OPL programs which are executed by the ILOG OPL Development Studio. Our proposal combines easy formulation of problems, expressed by means of a declarative logic language, with the efficiency of the ILOG System. Several experiments show the effectiveness of this approach.