Theory refinement and Natural Language Learning

  • Authors:
  • Hervé Déjean

  • Affiliations:
  • Universität Tübingen

  • Venue:
  • COLING '00 Proceedings of the 18th conference on Computational linguistics - Volume 1
  • Year:
  • 2000

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper presents a learning system for identifying syntactic structures. This system relies on the use of background knowledge and default values in order to build up an initial grammar and the use of theory refinement in order to improve this grammar. This combination provides a good machine learning framework for Natural Language Learning. We illustrate this point with the presentation of ALLiS, a learning system which generates a regular expression grammar of non-recursive phrases from bracketed corpora.