Matrix-Like Representation of Production Rules in AI Planning Problems

The paper presents the AI planning technology developed to study poorly formalized subjects domains, the knowledge of which is of a quantitative and qualitative character. The technology implements constraint programming paradigm and supports the subject domain model being open for operative modifications, which allows inclusion or exclusion of constraints, quality criteria, as well as setting the initial and goal states specified by subdefinite parameters. The originality of this work lies in the fact that a new type of constraints, namely smart table constraint of D-type, is proposed for representation and efficient processing of the production rules, with their processing being carried out by the authors’ methods of non-numerical constraints satisfaction, which gives the substantial gain in the performance against the conventional algorithms of table constraints propagation.