Valence Induction with a Head-Lexicalized PCFG

1998/05/05 by Carroll, Glenn, Rooth, Mats
#Computation and Language (cs.CL) #FOS: Computer and information sciences

paper · doi:10.48550/arxiv.cmp-lg/9805001

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.

Related

• Unsupervised Grammar Induction with Depth-bounded PCFG
• Inducing a Semantically Annotated Lexicon via EM-Based Clustering
• Can Subcategorisation Probabilities Help a Statistical Parser?
• The Return of Lexical Dependencies: Neural Lexicalized PCFGs
• PCFGs Can Do Better: Inducing Probabilistic Context-Free Grammars with Many Symbols