Work

Joint and conditional estimation of tagging and parsing models

Year: 2001
Type: conference-paper
Abstract: This paper compares two different ways of estimating statistical language models. Many statistical NLP tagging and parsing models are estimated by maximizing the (joint) likelihood of the fully-observed training data. However, since these applications only require the conditional probability distributions, these distributions can in principle be learnt by maximizing the conditional likelihood of the training data. Perhaps somewhat surprisingly, models estimated by maximizing the joint were superior to models estimated by maximizing the conditional, even though some of the latter models intuitively had access to "more information".
Author: Mark Johnson
Institution: Brown University
Language: English
Open Access status: bronze
OpenAlex ID: W2170469979
FWCI: 2.691
Cites: 19
Cited by: 53
Related to: 10