Valence Induction with a Head-Lexicalized PCFG
cmp-lg
2009-09-25 v1 Computation and Language
Abstract
This paper presents an experiment in learning valences (subcategorization frames) from a 50 million word text corpus, based on a lexicalized probabilistic context free grammar. Distributions are estimated using a modified EM algorithm. We evaluate the acquired lexicon both by comparison with a dictionary and by entropy measures. Results show that our model produces highly accurate frame distributions.
Cite
@article{arxiv.cmp-lg/9805001,
title = {Valence Induction with a Head-Lexicalized PCFG},
author = {Glenn Carroll and Mats Rooth},
journal= {arXiv preprint arXiv:cmp-lg/9805001},
year = {2009}
}
Comments
10 pages, 5 postscript figures