Multidimensional Predicates for Prolog

TL;DR

This paper introduces a library called mdp that extends Prolog with multidimensional predicates inspired by Korz, enabling clearer structuring of adaptive and context-aware systems through meta-programming techniques.

Contribution

It demonstrates how to integrate Korz's multidimensional model into Prolog, providing a flexible library for experimenting with adaptive system behaviors.

Findings

Enhanced structuring of Prolog programs with additional concerns

Facilitated experimentation with syntactical and semantical extensions

Showcased benefits in debugging, memoization, and adaptive GUIs

Abstract

In 2014, Ungar et al. proposed Korz, a new computational model for structuring adaptive (object-oriented) systems. Korz combines implicit parameters and multiple dispatch to structure the behavior of objects in a multidimensional space. Korz is a simple yet expressive model which does not require special programming techniques such as the Visitor or Strategy pattern to accommodate a system for emerging contextual requirements. We show how the ideas of Korz can be integrated in a Prolog system by extending its syntax and semantics with simple meta-programming techniques. We developed a library, called mdp (multidimensional predicates) which can be used to experiment with multidimensional Prolog systems. We highlight its benefits with numerous scenarios such as printing debugging information, memoization, object-oriented programming and adaptive GUIs. In particular, we point out that we can structure and extend Prolog programs with additional concerns in a clear and concise manner. We also demonstrate how Prolog's unique meta-programming capabilities allow for quick experimentation with syntactical and semantical enhancement of the new, multidimensional model. While there are many open concerns, such as efficiency and comprehensibility in the case of larger systems, we will see that we can use the leverage of mdp and Prolog to explore new horizons in the design of adaptive systems.