ACL 2007 General Chair Report

Firstly, I would like to thank all the people involved with organising ACL
this year. All of them carried out their roles to the full, were extremely
competent in everything they did, and were responsive and constructive in
discussions about any problems that cropped up.

I have thanked them by name in the General Chair Preface in the conference
proceedings (and I hope I have not inadvertently left anyone out). So in
this report I will only highlight a few issues that warrant some thought
for next year, or things that have been done differently than in previous
years.


1. Unreliability of email

One tutorial proposal got held up on a mail server and did not reach the
tutorials chair -- until he was prompted to look for it by the proposer
after the notification date. The problem was resolved by asking the
proposer to submit a new version, which was passed for comments to the two
reviewers who were most likely to have been given the proposal had it
originally been received before the deadline.

Lessons learned: either use START for tutorial (and perhaps workshop)
proposals, or stick with email submission but add a step to the process
"Acknowledgment of receipt" with a published date a couple of days after
the submission deadline, so that a proposer knows to expect an
acknowledgment and when it should arrive by.


2. Co-location of other, major workshops/conferences

As in previous years, decisions about co-location of major ACL SIG
workshops/conferences (e.g. IWPT, EMNLP) have to be taken before those for
standard workshops, since their schedules are on a longer timescale. The
same goes for workshops that include shared tasks; an example of this is
the SIGLEX-sponsored SemEval which submitted a workshop proposal early,
before the call for workshops proposals went out.

This year, the discussion about co-location of EMNLP-CoNLL was made harder
by the fact that the EMNLP organisers had talked to a number of people
connected with ACL 2007 at different points over the previous months, and
it took a while to piece together who had said what to whom. I suggest
that in future all such communication be directed through the workshops
chair (if they have been appointed by then) and the general conference
chair.

Another issue is that the EMNLP schedule has a strong interaction with the
main conference schedule since EMNLP wants to have their submission
deadline at least 1 day after the main conference notification date. This
dependency needs to be taken into account at an early stage in planning
the conference, otherwise the main conference notification date could end
up being too late for EMNLP to manage their submission process.

Another dependency: we set a recommended submission date for workshops to
be after the main conference notification date -- so that workshop PCs
would not have to deal with submissions that were also under consideration
for the main conference.

This year, the only other major conference in our field with a similar
reviewing period was AAAI; there were a small number of double submissions
which the PC chairs sorted out with their AAAI counterparts.


3. Birds-of-a-feather meetings

Birds-of-a-feather (BoF) meetings are short informal gatherings for
researchers with a common interest, often used by SIGs for their business
meetings, or for people in new research areas to plan future workshops.
There are several rooms available at ACL 2007 for this purpose, at
lunchtimes during the entire conference/workshop stretch. Priscilla
Rasmussen has been coordinating requests for BoF meetings and assigning
rooms.


4. The reviewing process

There were complaints from the authors of three submissions about the
quality of reviews they received. In two cases their complaints had some
substance. The authors of one submission compared the ACL reviewing
process to that of AAAI, in which authors have an opportunity to write a
response to reviewers' comments before acceptance/rejection decisions are
made. Future ACL conferences could perhaps consider adopting this model.


5. Workshops

There was an email discussion between the organisers about whether
participants should be allowed to register for overlapping workshops;
also, how often it happened that people registered for a 2-day workshop
and then instead mainly attended 1-day workshops in that 2 day period,
thus saving a few dollars in registration fees.

This is probably the first ACL with workshops lasting only half a day.
These slots were given either (i) to new, emerging areas which sounded
attractive and exciting, but for which it was not clear that there would
be sufficient submissions to fill a whole day, or (ii) to more established
areas where a similar workshop had run fairly recently so the amount of
new research might be limited. This idea needs to be evaluated to see
whether it should be tried again in future.


6. Submissions and final versions of papers

The Publications Chair will probably comment on the issues below, but I
want to flag them as well:

* Depending on how they are produced, PDFs can contain embedded
information that identifies the authors; this is obviously undesirable for
submissions. Perhaps in such cases the START system could be configured to
warn submitters and ask them to upload a new version without this
information?

* There were negotiations about the maximum numbers of pages for workshop
papers, at submission and for final versions. The previous limit of 8
pages is less valid now that so few hardcopies of workshop proceedings are
sold.

* The requirement for final versions of papers in US letter size caught
out many authors. Perhaps the STRRT system could again be customised to
check PDFs automatically for page size when they are uploaded?

John Carroll
May 2007