9. Grammar Induction through Implemented Machine Learning

  1. Alexander Clark and
  2. Shalom Lappin

Published Online: 21 MAR 2011

DOI: 10.1002/9781444390568.ch9

Linguistic Nativism and the Poverty of the Stimulus

Linguistic Nativism and the Poverty of the Stimulus

How to Cite

Clark, A. and Lappin, S. (2011) Grammar Induction through Implemented Machine Learning, in Linguistic Nativism and the Poverty of the Stimulus, Wiley-Blackwell, Oxford, UK. doi: 10.1002/9781444390568.ch9

Publication History

  1. Published Online: 21 MAR 2011
  2. Published Print: 7 JAN 2011

ISBN Information

Print ISBN: 9781405187848

Online ISBN: 9781444390568

SEARCH

Keywords:

  • grammar induction - through implemented machine learning;
  • formal computational learning theories - not offering actual learning systems for particular tasks;
  • machine learning, implementing algorithm - functions, for mapping elements, domain of input samples into values from a range of output;
  • samples of training set, annotated with classifications and structures - learning algorithm, output for test set;
  • unsupervised learning, training corpus - not annotated with structures or properties;
  • learning word classes, and NLP problems - applied to unsupervised learning, identification of parts of speech (POS);
  • Clark's method, how frequency and morphological information is incorporated - using texts prepared for MULTEXT-East project;
  • unsupervised learning, promising results in acquisition of morphology;
  • Klein and Manning's Expectation Maximization (EM) algorithm - selecting the most likely parse for a sentence;
  • unsupervised grammar induction systems - not as yet approximating human parsing abilities for domain-general text

Summary

This chapter contains sections titled:

  • Supervised Learning

  • Unsupervised Learning

  • Summary and Conclusions