Welcome to the first EACL newsletter of 2006. A lot has happened since the last newsletter. We had our 11th conference in Trento: the programme chairs and local organisation chairs have written reports on it for this newsletter. They ensured the success of the conference, and the EACL board are very grateful for their efforts. By the same token, the student board requested feedback on the student programme of EACL 2006. That feedback is now summarised on the EACL website (www.eacl.org). Please go and check it out. Once again, thanks are due to the student board for organising a very stimulating student session at EACL 2006.
The call for bids to host EACL 2009 has also now been issued. The call is included in this newsletter. Please consider hosting EACL 2009. If you have any questions about what's involved, then please email me (alex@inf.ed.ac.uk; subject line EACL 2009).
In addition, we have a report from Ido Dagan on the last PASCAL meeting, and from Claire Gardent on TALN. In spite of the increase in the number of conferences and workshops in computational linguistics, many people are currently troubled by the low acceptance rates for ACL/NAACL/EACL conferences. Diana McCarthy and Shuly Wintner, the programme chairs for EACL 2006, have written an article about this issue, and I hope it will provoke thought and discussion.
On that note, in the run-up to EACL 2009, the EACL board wants to think hard about how we should react to this trend towards extremely low acceptance rates, and in particular what can be done to encourage more submissions of thought-provoking and perhaps controversial research. We are open to suggestions, which can be considered by the future programme chair for EACL 2009. If you have opinions about this, please forward them to me (alex@inf.ed.ac.uk; subject line EACL PROGRAMME).
Priscilla's report gives details about EACL membership, and it indicates how current European activity in computational linguistics is vibrant and thriving. In view of this, it becomes vital that EACL as an organisation maintains some corporate memory: Anette Frank has written a report on the EACL web archive which is one attempt to address this problem. She has also reported on EACL's recent student sponsorships. Finally, as always, Sebastian Pado has supplied us with a Calendar of future events. Sebastian steps down from the EACL student board in June. I would like to thank him for all his efforts over the past few years, especially on maintaining the EACL website. He has been fantastic. He will be replaced by Nuria Bertomeu from 1st July. We'll tell you more about her in the next newsletter!
In April, the 11th conference of the European Chapter of the ACL took place in Trento. As well over 300 participants have witnessed, Trento is a beautiful historical city on the border of north and south, exhibiting not only many Italian aspects, but also some Germanic ones, as exemplified during the banquet for instance! In the week of the conference, even the weather exhibited southern and northern aspects, ranging from a nice spring day on the Sunday when everybody arrived, to some rain and some snow (if you looked up at the mountains surrounding Trento in all directions), and again lots of sunshine at the end of the week.
Many volunteers helped to make EACL 2006 successful. Thank you to all of you! Thank you to Diana McCarthy and Shuly Wintner, the main program chairs, and thank you to the chairs of the local organization committee, Bernardo Magnini and Alberto Lavelli.
According to the ACL/EACL schedule, the next major ACL event in Europe will be the ACL 2007 meeting in Prague. In 2009, we will organize the 12th conference of the European Chapter. You must have seen the call for bids for the 2009 EACL conference, it was published just a few weeks ago. Please consider bidding for this conference!
Twenty-three years after the first EACL conference in Pisa, EACL took place again in Italy, at the beginning of April in Trento.
Although strongly international, with 35 countries represented, the conference retained a distinctive European flavor: 73% of the participants (316 in total) were from Europe (23 countries), 16% from America, 10% from Australasia and 1% from Africa.
As usual, the event consisted of a main conference, a student research workshop, posters, demos, 12 workshops and 4 tutorials. For the main conference, the acceptance rate was about 20%, while posters had an acceptance rate of 38%. Demos had a 57% acceptance rate, whereas the student research workshop adopted a selective 27% acceptance rate.
The main conference for EACL 2006 had a 46% increase in submitted papers compared to the 10th EACL which was held in 2003. This year we had 264 papers submitted from 35 countries in five continents. The large number of papers submitted, and the broad geographical basis, demonstrate the growth in our field and considerable interest in EACL as a truly international conference. All papers were read by a minimum of three people, with discussions where there were differences of opinion. The final selections were made at a one and a half-day program committee meeting in Brighton. Of the 264 submissions, 52 papers were accepted giving an acceptance rate of 19.7%.
The conference also contained tutorials, workshops, posters, demos and a student workshop. Keynote addresses were delivered by Kevin Knight, of the Information Sciences Institute, University of Southern California, and Alfonso Caramazza, of the Cognitive Neuropsychology Laboratory, Harvard University. We would like to thank the invited speakers, the 240 reviewers, the 10 area chairs and all authors and participants for helping to make this conference a great success.
The European Chapter of the Association for Computational Linguistics (EACL) hereby invites proposals to host the conference of the European Chapter of ACL in 2009. This conference takes place every 3 years. The conference in 2009 will be the 12th meeting of the EACL.
At this time, we seek draft proposals from prospective bidders. Based on an evaluation of the draft proposals received, promising bidders will be asked to provide additional information on which basis final selection will be made.
The EACL Board will select the general chair for the conference, the programme committee chair, and all other chairs (e.g., tutorial chair, workshop chair, sponsorship chair etc) except for the local arrangements chair.
Draft proposals should identify a local arrangements chair, who will be responsible for activities such as arranging meeting rooms, equipment, refreshments, accommodation, on-site registration, participant internet access, security for equipment, the reception, the banquet, and working with the other chairs and the EACL Board to develop the budget and registration materials. Achieving budget estimates will, in particular, involve liaising with the sponsorships chair, who will be chosen by the EACL board.
Draft proposals are due on December 15, 2006. They will be evaluated by the EACL Board by February 1, 2007. Promising bidders will then be contacted and asked to provide further information due by May 13, 2007. The final bid will be chosen in July 2007.The conference should take place during April 2009, avoiding national holidays of the EACL countries (in 2009, western Easter is April 10--13, orthodox Easter is April 17--19, Jewish passover is April 9-16). Draft proposals should include information on:
Proposals will be evaluated in relation to a number of site selection criteria (unordered):
The ACL conference handbook and more about ACL policies may be found at: http://www.aclweb.org/archive.
Successful sample bids for previous conferences may be found at: http://www.aclweb.org/archive/ bids.html.
Please send draft proposals electronically to the EACL Chair-elect:
Alex LascaridesYou can also attach your draft proposal as a pdf attachment in email.
The Second PASCAL Recognising Textual Entailment Challenge (RTE-2) Workshop was held in the beautiful venue of Venice, Italy, on April 10 (immediately following EACL), as part of the Second PASCAL Challenges Workshop. The participation rate this year shows growing interest in textual entailment: 23 groups, including 12 from Europe, 10 from North America, and one from Australia, submitted to the second challenge (35% growth), including 13 groups participating for the first time. 17 out of the 23 groups presented their systems at the workshop, and overall about 50 people attended the workshop.
The workshop brought together researchers from various NLP communities, such as question answering, summarization, information extraction and semantic modeling, including academia and industry. It provided an opportunity to share different approaches to textual entailment, present novel ideas and discuss common issues. The results achieved this year show considerable improvement comparing to RTE-1, and the diversity of new approaches and research directions are very promising for further research. Overall, it seems that both learning-based and logic-based approaches are gradually progressing to address semantic inference problems in a generic robust manner.
The workshop was devoted mainly to system presentations, including 6 long presentations and 11 short presentations. In addition, two panel sessions were held, one on "Perspectives on Textual Entailment" (panelists: Sanda Harabagiu and Ido Dagan), and the other on "Goals and Dataset Proposals for RTE-3" (contributors: Lisa Ferro, Vasile Rus, Anselmo Peas, Johan Bos and Nicola Stokes).
The Challenge was co-organized by Bar-Ilan University (Israel, coordinator), CELCT (Italy), ITC-irst (Italy), Microsoft Research (USA) and MITRE (USA).
TALN'06 (Traitement automatique des langues naturelles/Natural language processing) took place in Leuven (Belgium) from 10th--13th April, 2006. Initiated by Philippe Blache in 1994 and taking place on a yearly basis, TALN has become the french speaking conference for computational linguistics and natural language processing. Each year it gathers between 100 and 300 researchers and industrials working in those fields (300 when the conference is joined with the french speaking conference for speech processing ie JEP, Journees d'Etude sur la Parole).
As usual, this year TALN featured a rich program consisting of 34 papers, 30 posters and 9 demos. The program also included 11 student papers, one tutorial, one workshop and two invited talks. The first invited talk was given by Gertjan van Noord (U. of Groningen, The Netherlands) who presented work on robust parsing for Dutch, a topic highly relevant to the French community which like many others is still working at developing tools and resources that have long become standard in the english based commmunity. The second invited talk by Thierry Fontenelle (Microsoft) focused on the linguistic correction task in the Microsoft system.
A recent additional feature of the conference is that it has also become a forum for presentation of national evaluation campaigns. Thus the fourth day was devoted to the presentation and discussion of various evaluation campaign results and of relevant new initiatives such as the creation by the CNRS (French National Research Center) of five Centre de Ressources Numeriques (Digitised resource centers), several of them should contribute to the development and promotion of freely available linguistic tools and resources for the French language.
Previous editions of TALN have taken place in Dourdan, Fes (Maroco), Batz-sur-mer, Nancy, Tours, Lausanne, Cargese (Corsica), Paris, Grenoble et Marseilles (3 times). Next year TALN is already being planned: it will take place in sunny Toulouse organised jointly by Philippe Muller (IRIT) and Nabil Hathout (ERSS).
EACL 2006 had an acceptance rate of 19.7%. Many in our community feel that this rate is too low. This concern was made explicit in a recent Computational Linguistics article by Ken Church. We would like to address these concerns and make a few suggestions for future editions of EACL.
First, while the decisions on acceptance were our own (with the help of the 10-member Program Committee), they reflect the recommendations of 240 reviewers. When there is a consensus among reviewers that a paper is either mediocre or suffers from serious flaws, we as program chairs do not feel it should be accepted. If more papers are to be accepted, they must have better reviews. We cannot be sure that we did not miss some good works; but we did do our best to spot promising directions, even if the work described in the paper was preliminary.
Second, while acceptance rate indeed dropped some seven percent points from the last EACL, the absolute number of accepted papers was up from 48 to 52.
Third, acceptance rates of around 20% are common in the best conferences in other areas of computer science, such as theory (e.g., SToC or FoCS). In areas such as computer vision or bioinformatics, even lower rates are common.
Still, we believe that there are ways to allow more papers into the program of leading conferences such as EACL. One solution is to provide a channel for presenting more preliminary, but highly innovative and original work. This could be in the form of a "research notes" track, possibly with a later deadline than the main session (and, subsequently, fewer pages in the Proceedings). Another direction could be the poster session: this could be dedicated to works which may be of limited interest to the entire community, but which may be attractive to smaller groups, especially where the authors can benefit from the more interactive feedback that posters provide. In addition, the Program Committee could be given the prerogative to move submissions from the main session to either of these tracks, if it feels that a paper is more appropriate there. Whatever the solution, it is up to us as a community to implement: as reviewers, as area chairs, as program chairs but first and foremost, as authors.
We always welcome suggestions about how these issues might be addressed by structuring the EACL conference programme in particular ways. If you have views on this, then please email Alex Lascarides (alex@inf.ed.ac.uk). Do *not* email Shuly Wintner or Diana McCarthy! Your suggestions will be summarised in a future edition of the newsletter, and also discussed at future EACL board meetings.
ACL membership among countries that are in the EACL area is currently down compared with 2005 membership by 24%: last year's total membership was 403 while this year it's 308 (as of 9th May). However, the spread of membership geographically is very wide, with members coming from 36 countries. There are 272 members in EU countries, 14 from the middle East, 13 from Eastern Europe and 9 from other countries (i.e,. South Africa and Switzerland). One interesting fact is that the proportion of student membership for this year so far is down by a third from 2005 (from 25% to 16%). EACL would like to monitor this, in order to assess whether PhD recruitment in computational linguistics is a problem in Europe, and if so whether it's more pronounced in Europe than elsewhere. However, given that many people become members through their conference attendance and we have two upcoming events (HLT/NAACL 2006 in New York and COLING/ACL 2006 in Sydney), it is highly likely that we will hit the 2005 membership totals or exceed them.
The EACL Executive Board has decided to set up a private web page from www.eacl.org for EACL officers, with the aim of collecting and continuously maintaining a "collective memory" of policies and practices, and a repository of reports and documents. This ensures the information is readily available with changing composition of the board over time. The pages are maintained by the secretary and the Student Board.
We are currently composing an archive of previous EACL officers. The current list is mirrored at the public EACL web site. We invite previous members of the EACL Board to help us complete this list, by sending a short email to secretary@eacl.org or students@eacl.org, indicating their function and time of service.
This year, EACL is offering substantial financial support for ESSLLI 2006, valuing 3000 Euros in total. Preference will be given to students of Eastern European countries. For details of application see: http://esslli2006.lcc.uma.es/give-page.php?id=14
The EACL has also funded five students from the geographical area of the EACL who had papers accepted at the EACL 2006 conference (in the main session, the student research workshop, the poster/demo session, or co-located workshops).