Unsupervised approaches receive considerably growing attention in NLP in the
last years and dependency parsing is not an exception. The motivation is simple:
Such parsers does not need any manually annotated treebanks and might be able to
parse texts in whatever language and domain. Although the parsing
quality is still
far below the quality of supervised parsers, we can observe their substantial
improvements at each NLP conference.
I will present the most popular statistical approaches that are used
for automatic induction of grammar from text. In the second part of the talk,
I will describe my own method, which is based on Gibbs sampling procedure and
which proved to compete with state-of-the-art systems.
default – not confidential
Creator: Common Account
Created: 6/24/12 1:26 PM
Modifier: Common Account
Modified: 6/24/12 1:26 PM
Content, Design & Functionality: ÚFAL, 2006–2016.
Page generated: Mon Oct 15 18:25:34 CEST 2018