Natural language research announcements

["Alexandre Klementiev" <klementi AT uiuc.edu>]

Dear all,

The location of the talk has changed: it will take place in Siebel 3403 (next to our usual room).

Alex.

On Sat, May 17, 2008 at 3:14 PM, Alexandre Klementiev <klementi AT uiuc.edu> wrote:

Dear faculty and students,

Gabriele Antonio Musillo will give a talk (details below) at the AIIS seminar this Thursday.  If you would like to meet with Gabriele personally, please let me know.

Thank you,
Alex

Title: Unlexicalised Hidden Variable Models of Split Dependency Grammars 
Speaker: Gabriele Antonio Musillo, University of Geneva and MIT

Date: May 22, 4:00pm
Location: Siebel 3405 

Abstract: 

This talk presents transforms of split dependency grammars into unlexicalised context-free grammars annotated with hidden symbols, that make learning and decoding algorithms for probabilistic context-free grammars available to projective dependency grammars. The Inside-Outside algorithm is used to estimate the hidden variables decorating such transforms. Our best unlexicalised grammar achieves an accuracy of 88% on the Penn Treebank data set, that represents a 50% reduction in error over previously published results on unlexicalised dependency parsing. Such performance shows that our unlexicalised models are able to capture both lexical and structural information that is relevant to parsing accuracy and suggests that we should reassess the relevance of massive lexicalisation to dependency parsing.

Bio:

Gabriele Antonio Musillo is a PhD student in Computer Science and Linguistics at the University of Geneva, Switzerland, under the supervision of Prof. Paola Merlo. His main research interests lie in statistical parsing of natural language and machine learning of natural language grammars. He is currently a visiting student at MIT/CSAIL hosted by Prof. Michael Collins.