2012Q3 Reports: EMNLP 2011

From Admin Wiki
Jump to: navigation, search

Report on EMNLP 2011

EMNLP 2011 --- the Conference on Empirical Methods in Natural Language Processing, a conference organised annually by SIGDAT, the Association for Computational Linguistics' special interest group on linguistic data and corpus-based approaches to NLP --- took place from July 27th to 29th 2011 at the John McIntyre Conference Centre, Edinburgh, UK. This EMNLP was, for the first time, not only a stand alone conference, but also an anchor conference to several workshops, that were held on July 30th and July 31st at the Informatics Forum, Edinburgh. This year's conference continued the successful growing trend of previous years, attracting the largest number of papers to date for EMNLP and requiring a large organisational effort. The conference was preceded by a very interesting colocated event sponsored by Google and the Scottish and Informatics Computer Science Alliance: an Intense Summer School on Hadoop and Natural Language Processing.


This was the first time EMNLP had a general chair, Paola Merlo, Geneva, Switzerland. The scientific programme chairs were Regina Barzilay, MIT, US, and Mark Johnson, Macquarie University, Australia. Publication chair was Wanxiang Che, Harbin Institute of Technology, China. Workshop chair was Marie Candito, Paris 7, France. Local arrangements were being organised by Bonnie Webber and Miles Osborne, University of Edinburgh.

The list of 20 area chairs can be found at http://conferences.inf.ed.ac.uk/emnlp2011/area-chairs.html

EMNLP received 628 submissions (not counting papers that were withdrawn or rejected without review). We were able to accept 95 papers as talks and an additional 54 submissions as posters. The reviewing process was a two-round review with author response period.

This conference was innovative in several ways. All three morning in the conference were devoted to plenary sessions; these were used to highlight a diverse set of papers of interest to the entire EMNLP audience. We hope this will help counter the disciplinary fragmentation that some of us feel the standard multi-track conference structure encourages and were in general appreciated by the participants.

For the first time, submitted papers could be optionally accompanied by up to 10MB of supplementary material, which could consist of data, code, and text. Papers could reference the supplementary material in much the same way a paper might refer to software or a tech report available from the authors' web site (albeit without revealing the authors' identities). Reviewers were encouraged but not required to view the supplementary material. We did include these unreviewed materials in the proceedings. Roughly 20% of the papers took advantage of this option.

A major challenge this year concerned undisclosed double submissions and plagiarism (especially self-plagiarism) involving papers accepted at other international conferences. We believe this is an issue that must addressed by the broader research community.

The conference attendance largely exceeded the 400 people, and the weather was uncharacteristically dry and sunny.

Paola Merlo General Chair for EMNLP 2011