December 9: Conditional Random Fields

We will read the following paper:

J. Lafferty, A. McCallum and F. Pereira. Conditional Random Fields: Probabilistic models for segmenting and labeling sequence data. Proc. 18th Int. Conf. on Machine Learning, 2001.
http://www.cis.upenn.edu/~pereira/papers/crf.pdf

In addition, if time permits, we will reread the previous CRF paper:

Charles Sutton, Andres McCallum. An Introduction to Conditional Random Fields for Relational Learning.
http://www.cs.umass.edu/~mccallum/papers/crf-tutorial.pdf

November 25: cancelled

Martin, Matthias and Laura are teaching a ph.d. course in statistics on November 25, so we decided to cancel the meeting.

November 11: Conditional Random Fields

In continuation of the ML-reading group we will discuss the theory behind and the usage of Conditional Random Fields.

If anyone knows of a NLP-article using CRF (and explaining it's usage) please put a link to the article here (or mail it to Martin).

October 28: Galley et al 2004 and 2006

We read about the syntax-based SMT system that was referred to from the paper ”Two Langauges are Better than One (for Syntactic Parsing)”, ie, the following two papers:

Galley, Hopkins, Knight, Marcu, "What’s in a translation rule?",Proc. HLT-NAACL 2004.

http://aclweb.org/anthology-new/N/N04/N04-1035.pdf

Galley, Graehl, Knight, Marcu, DeNeefe, Wang, Thayer,

"Scalable Inference and Training of Context-Rich Syntactic Translation Models", Proc. ACL 2006.

http://aclweb.org/anthology-new/P/P06/P06-1121.pdf

All the best,
Matthias

PS. I am unable to attend the reading group myself because of a funding meeting. :-(

October 14: Anders' work on semi-supervised dependency parsing

On October 14, Anders will present his work on semi-supervised dependency parsing.

NLP reading group September 30

The reading group on September 30 (14.30-15.30) is devoted to a discussion of the research strategy in the CBS NLP group. Andreas and Anders, feel free to join the discussion or leave early as you please.

Matthias

September 16: Haghighi, Blitzer, DeNero and Klein

Kære alle

Den 16. september læser vi:

Better Word Alignments with Supervised ITG Models af Aria Haghighi; John Blitzer; John DeNero; Dan Klein
http://john.blitzer.com/papers/itg_acl09.pdf

Venlige hilsner
Matthias

September 2: Burkett and Klein.

Kære alle

I NLP-læsegruppen på onsdag (2.9. 2009) vil jeg præsentere følgende paper:

• Two Languages are Better than One (for Syntactic Parsing) by David Burkett and Dan Klein. http://www.eecs.berkeley.edu/~dburkett/papers/burkett08-joint_parsing.pdf

Hvis I ikke allerede har oprettet jer som læsere (Followers, se menu i højre margen) af NLP-læsegruppe bloggen, må I meget gerne gøre det. Blog-adressen er:

http://cbs-natural-language-processing.blogspot.com/

Venlige hilsner
Matthias

August 5: presentation by Michael Carl

Dear all,

Michael Carl has volunteered to present the following two papers in the NLP reading group meeting on August 5 (from 14-15.30 in 2Ø.091).

All the best,
Matthias

---

On 2009-07-28 14:08, Michael Carl wrote:
I would like to present and discuss a forthcoming journal paper in which we propose a methodology for gathering and analysing translator's activity data. We compare three human translations of an English text into Danish and look at a number of correction patterns and a fixation pause in more detail. The paper suggests that based on a better and more formalised way of understanding the underlying human processes, new technologies and novel ways of using existing technology could emerge that tighten the integration of automated translation aides into translation workflow.

Another forthcoming paper from Philipp Koehn gives another view on interactive MT.

Purpose

The reading group is intended as a forum where researchers and their students can keep up-to-date on developments in natural language processing, and discuss books, papers, and software packages with state-of-the-art approaches to NLP.

All interested researchers and students, from all institutions and all fields, are welcome in the reading group.