Parsing and Generation with Tabulation and Compilation

TL;DR

This paper introduces a data-driven compilation method for logic programming that enhances flexibility and efficiency in parsing and generation by integrating control techniques like accessibility checks and last-call optimization.

Contribution

It presents a novel data-driven constraint transformation and compilation approach that generalizes standard parsing and generation techniques in logic programming.

Findings

Improved parsing and generation efficiency

Unified framework for accessibility and optimization

Enhanced adaptability to diverse contexts

Abstract

The standard tabulation techniques for logic programming presuppose fixed order of computation. Some data-driven control should be introduced in order to deal with diverse contexts. The present paper describes a data-driven method of constraint transformation with a sort of compilation which subsumes accessibility check and last-call optimization, which characterize standard natural-language parsing techniques, semantic-head-driven generation, etc.