Practical experiments with regular approximation of context-free languages

TL;DR

This paper explores various methods for approximating context-free languages with finite automata, including new approaches, and evaluates their effectiveness on spoken-language input in speech recognition tasks.

Contribution

It introduces new regular approximation methods and refines existing ones, providing practical experimental evaluation on speech recognition data.

Findings

New approximation methods demonstrated effectiveness

Refined existing methods improve approximation quality

Practical filtering of speech hypotheses achieved

Abstract

Several methods are discussed that construct a finite automaton given a context-free grammar, including both methods that lead to subsets and those that lead to supersets of the original context-free language. Some of these methods of regular approximation are new, and some others are presented here in a more refined form with respect to existing literature. Practical experiments with the different methods of regular approximation are performed for spoken-language input: hypotheses from a speech recognizer are filtered through a finite automaton.