2010Q3 Reports: SIGDAT (SUBMITTED)
SIGDAT - February 2010 Report
SIGDAT is ACL's special interest group for linguistic data and corpus-based approaches to NLP.
In 2009, SIGDAT organized its annual Conference on Empirical Methods in Natural Language Processing in Singapore on August 6-7, immediately following ACL 2009. Philipp Koehn and Rada Mihalcea were the co-program chairs, Haizhou Li was local arrangements chair and David Chiang was the publications chair. There were 15 area chairs. The conference received 478 full paper submissions (a record) and approximately 100 papers were ultimately accepted for oral presentation and 58 submissions were accepted for poster presentation. Attendance exceeded 300 people. EMNLP 2009 was held in 4 parallel sessions and the proceedings were nearly 1500 pages long. On all dimensions EMNLP and SIGDAT had a record year in 2009, with EMNLP exceeding the size of most pre-2000 ACL conferences and similar to recent EACL and NAACL-HLT main conferences in size.
The SIGDAT committee has voted to hold EMNLP 2010 in the northeast US during late October 2010. Based on the longstanding democratic traditions of SIGDAT governance, a survey of the entire SIGDAT membership was conducted to identify the preferred location, duration and dates of the meeting, as well as to achieve an initial indication of likely paper submission and attendance. Over 260 SIGDAT members responded, including 53% from North America and roughly 25% from Europe and Asia respectively. There was a strong collective preference for holding the meeting in the Boston/Cambridge area over a Fri/Sat/Sun weekend with 2-3 parallel sessions (over several other location and scheduling options). There was less consensus regarding the preferred weekend but sufficient rank ordering to help guide our search for a venue with availability. Interestingly 98% of respondents (over 255) said that they planned to submit papers to EMNLP 2010, 99% of those planned to attend if they had a paper/poster accepted and over 80% said that it was likely they would attend even if they *didn't* have a paper or poster accepted. These numbers indicate that attendance of at least 225 is likely, helping SIGDAT with its venue choice and budget planning. Based on the traditionally accelerated SIGDAT reviewing schedule a late June or early July submission deadline is anticipated, facilitating submission of late-breaking work and minimizing conflict with other 2010 conferences. The SIGDAT organizers will seek ways to facilitate attendance of participants requiring US visas under this schedule.
The SIGDAT committee has also voted to hold EMNLP 2011 as an anchor event in late summer 2011 in Europe, to help compensate for a general lack of NLP/CL conferences in 2011 (with no COLING or EACL and one potentially huge ACL meeting in Portland in early summer). A three-day EMNLP conference plus 1 day of associated workshops is the committee recommendation, with potential direct collaboration with other larger ACL SIGs under discussion. We also request to participate in the standard coordinated joint proposal submission process for smaller workshops with the main ACL conference, given that ACL 2011 by itself will most likely not have capacity to support all of the workshops normally held over 3-4 CL/NLP conferences in a more typical year. SIGDAT believes that this special EMNLP anchor event will help provide load balancing benefiting the broader ACL community in a year with many fewer than normal potential paper presentation slots. A general chair appointment is currently under review to take on this significant organizational role.
Under normal continental rotation, EMNLP 2012 would normally be held in Asia, and collocation with ACL 2012 in Asia would seem to have benefits. CoNLL has expressed some interest in a potential joint meeting in 2012. Given the less immediate time sensitivity of this decision, it may be deferred until after new SIGDAT elections are held (infrastructure for doing so is nearly in place) or made via direct vote of SIGDAT's membership, as part of a series of forward looking questions for the SIG.
Overall, SIGDAT has had a continuing history of organizational success, with three consecutive years of 400-500 submissions and 230-300+ attendees, suggesting the viability of SIGDAT continuing to organize off-phase stand-alone EMNLP meetings in the future, especially when held in North America or Europe, as well as its ability to help add critical mass to other conferences when held in collocation.