APRIL 22, 2007

  1. 2007 HLT-NAACL
  2. Calls for bids for 2009
  3. The 2008 meeting, joint with ACL in Columbus
  4. Advisory board
  5. Treasurer's report
  6. SLT
  7. Summer Schools
  8. Miscellaneous topics
[1] 2007 HLT-NAACL

Status of the conference as of the Exec meeting

437 (32 workshop-only) had registered, with an estimate of 475 eventual registrations, which is a decline from last year's 689. (The number of submissions seemed about on par with those last year; the number of workshops was down whereas ACL was oversubscribed with workshop submissions.) Business Manager Priscilla Rasmussen was therefore pulling back from a 500-person break-even budget to a 400-person break-even budget (e.g., reduce internet service from a $5000 to a $3000 plan). On the other hand, Google provided an unexpectedly large amount of $2500 in the previous week. The general sense was that the early timing of the conference, rather than the location, was the probable cause of the decline. There was discussion of whether May or October would be better, with some thinking that a 2009 ACL in August in India would imply a 2009 NAACL in May. Other possibilities were the fact that this year's timing was back-to-back with ICASSP, or that grant funding might not have been finalized this early, suggesting late October as a possibility (early October is not good for government folks, since they wouldn't have a travel budget then). Scheduling for 2009 needs to take into account EACL 09, thought to be either April 27-May 1 or March 27-April 1.

A GALE meeting this year caused the full-papers submission deadline for to HLT-NAACL be pushed later.

Short papers

The NLP Program Chair Matthew Stone noted that short papers seem to need different treatment, since they seem to attract different audiences and/or interact with other conferences' deadlines in interesting ways, and this has potential implications for the structure of the conference. There was a good representation of IR papers among the full (notification deadline pre-SIGIR), but not the short, papers; the situation was the exact reverse for speech (deadlines in October).

Conference organization: observations and proposals

Matthew Stone found the triumvirate organization unwieldy, both because of the different loads mentioned above, and because of the overhead triggered (since more people were involved in decision making). A checklist of advice would include having a meeting by phone once a week, and having certain tasks (such as the creation of area sub-committees, where only one person might have the relevant expertise) have the decision-making delegated to a single person, although in practice, one person often seems to become "the leader" anyway. Strong words were used to express the opinion that untenured people should not be asked to serve as program co-chairs, and that candidate co-chairs need to be informed of the workload up-front.

One suggestion was to have the NLP person be the first among equals (is the NLP person generally more involved in the conference itself as it forms part of the basis of his/her own personal research community?). Or how about two NLP co-chairs? But perhaps more important than one person being in charge is figuring out a way to allow people to contribute easily; part of issue is people coordinating when it is they need to do what. Specific problems: getting reviewers for speech during Interspeech; determining the *two* invited speakers (which in part means we need to invite PC Chairs much earlier). On a related note, perhaps no main PC chair is needed for speech, given that there were almost no full-paper submissions? Indeed, should the area split be handled through area chairs/senior PC members rather than at the program chair level, with the advisory board asked for advice for the selection of senior PC members when necessary? It was strongly suggested that the senior PC should not have to handle the short papers, especially since the reviewing is supposed to be different for these papers and the same reviewers aren't likely to switch evaluation criteria.

Also, demos and publications didn't make sense as an 3-area triumvirate. That is, we could have had three people, but they needn't have been from different areas. Indeed, publications is three people's worth of work, and could use extra secretarial support as well; whereas for sponsorships and exhibits, a triumvirate simply was not set up.

It would be beneficial to have a list of recommended START-system settings (such as were used by a recent EMNLP), instructions for how to carry over the full paper "apparatus" to the short-paper reviewing; and previous review forms; this could perhaps all enter the Conference Handbook or the archives. It might also be useful to have NAACL offer support, e.g., for a graduate-student in HLT to be able to check for double submissions (a task that it does not seem an administrative assistant could handle). We designated a sub-committee for thinking about reviewing procedures and software: Regina Barzilay, Jennifer Chu-Carroll, Lillian Lee, and the three PC chairs.

Handing down past experience

General chair Candy Sidner has requested "lessons learned" summaries from various organizers [These reports have been placed in the 2007 section of the NAACL archives], and suggests that there be an NAACL version of the ACL Conference Handbook, and that lessons learned be passed along. [Note that prior conference reports are available at the NAACL archives.] This suggestion was later refined by ACL Business Manager Priscilla Rasmussen to a "supplemental handbook to the ACL Handbook for NAACL meetings", which would perhaps highlight deviations from ACL policy and NAACL-specific items, and would perhaps specifically mention defaulting to ACL policy in the relevant places. It was also suggested that organizers at all levels be reminded to check the NAACL archives.

Visa issues for conference attendees.

One Italian prospective conference attendee was denied a visa, perhaps because of a citizenship issue, but despite supporting letters from the conference organizers. The organizers contacted the office of a U.S. Senator from New York State, to no avail. Visa problems have been an issue for the AAAI and for previous NAACL speakers, as well. Upon review, it does not appear that different timing or a different letter would have solved the problem. There was some speculation that visa problems have actually eased recently.

Suggestions for the future for conference attendees: inform authors (via the conference webpage and other means) that that these problems can arise, and suggest that although the organization can try to help, authors need to: act early; try to identify alternate presenters who could give the paper in their stead; and inform senior PC members (who can inform session chairs) if they haven't received a visa within one or two weeks of presentation date. There was some discussion of whether having reserved (just-in-case) papers makes sense: on the positive side, this means that unexpectedly open slots (visa problems, illnesses) can be filled; on the minus side, this may force reserve-slot authors to incur travel costs perhaps unnecessarily, and might send an inadvertent signal that people sometimes choose not to attend the conference.

Suggestions for the organization: check into how companies get expedited visa handling; consider collaborating with other scientific organizations; bring the issue to the attention of the ACL, who may be the body for whom the ultimate responsibility should lie.

(return to top of minutes)

[2] Calls for bids for 2009

Some possible locations and groups to organize were discussed.

(return to top of minutes)

[3] The 2008 meeting, joint with ACL in Columbus, with a Coordinating Committee involved.

(See also the conference-organization notes above.)

We discussed the question, "How much do we want ACL 2008 to feel like an NAACL?" The aspects we discussed were: 1) outreach to speech/IR; and 2) the short-paper track. The claim was made that people find the ACL poster sessions less interesting than the HLT late-breaking papers session. The NLP program chair for this year's NAACL, Matthew Stone, felt that the triumvirate structure (speech, IR, NLP) was unwieldy. (See also discussion of the short-papers track in the 2007 meeting, described above.) At any rate, if there are doubts in the NAACL's mind regarding the triumvirate structure, it's not clear that we should ask the ACL to conform to it. The group agreed that we should support having short papers at the meeting.

Some discussion of whether the 2008 meeting should have the HLT name ensued. The upshot seemed to be that the group didn't have strong feelings about the matter.

Proposals for ACL 2008 follow

Some suggested having two PC chairs, and having a separate short-papers committee with two chairs also; but there was strong opposition to this, with the counter-suggestion to keep the Speech PC chair. The eventual recommendation was to have the IR and the speech PC chair handle both long and short submissions, to have an NLP PC chair handle long submissions, and another NLP PC chair handle short submissions, with the reminder that the advisory board serves as a resource, and the clarification that short papers go in a separate volume, and the suggestion that short-paper and demo reviewing could be combined.

Getting the names of the senior PC on the call for papers would be good, since this helps to show the areas that we're interested in.

Policy decision: PC chairs, up to three local chairs, and conference chairs receive free registration in recognition of their contributions.

(return to top of minutes)

[4] Advisory board

As of the meeting, Jeff Reynar had agreed to join, and Donna Harman was working on getting an official response from SIGIR regarding an official rep. There was some discussion of how much input the advisory board should "officially" have into the ACL meeting next year, of how different areas should be balanced with respect to representation on the advisory board, and of whether non-North-American entities, such as ELRA or AFNLP (who runs IJCNLP) could be contacted, although ACL Treasurer Kathy McCoy noted that there can be trouble with what kinds of entities can receive funds of various kinds of origins. The final decision was to stick with a North American board for now, but suggest that the ACL think about expansion beyond this.

(return to top of minutes)

[5] Treasurer's report

The report is archived [see the NAACL archives]. At the time, HLT-NAACL 2004 accounting had been finished after about two Sundays' worth of time, but ACL 2005 accounting had not been completed by the ACL Treasurer yet.

There was a quick reminder that the North American Computational Linguistics Olympiad had not put an acknowledgment of NAACL funding on its webpage, but this was presumably an oversight.

Fund-raising possibilities discussed: recruiting sessions, awards, and/or student luncheons that companies might want to sponsor; or, bundled "sponsorship packages" to make sure the funds can also be used towards general expenses (if companies are willing to buy into such packages; there was some doubt in this regard). NAACL Chair Owen Rambow suggested that in order to not compete with ACL and ACL sub-organizations, fundraising should be coordinated across the "*ACL"s, with perhaps two-year coordinating positions to allow for the cultivation of personal relationships with industry contacts. One committee possibility would be a global position, plus a chapter position, plus a local position. Exec Member Bill Dolan and Owen Rambow decided to form part of a committee to deal with these issues.

Then, discussion turned to ways to use our surplus. One option that was discussed was resource creation, such as software and data repositories or archives. (However, it was noted that it could be difficult for industry folks to get licenses, and that licensing issues are important with respect to good relations with industry. Thus, a repository would probably need to articulate different levels of permissions.) It was also noted that the patent and trade office looks at or uses the ACL anthology. There was some sentiment that we shouldn't fund items for which other funding sources exist, and that we shouldn't put out an open proposal call. Another option was to alter the costs of conference attendance.

(return to top of minutes)

[6] SLT

We discussed IEEE's new Spoken Language Technology workshop, SLT, and IEEE's "semantic computing" initiative. Board Member Mary Harper suggested strengthening ties with ICASSP, a community whose use of statistical techniques forms a natural bridge, and ICSLP, which is more human-centered than the typical NAACL meeting tends to be. Such strengthening might consist of asking for recommendations for advisory-board members, co-sponsoring conferences, co-announcing things in newsletters, or joint discounts. Perhaps a similar arrangement could be made with SIGIR. In both cases, the "other" organization could designate an advisory-board member as their representative.

(return to top of minutes)

[7] Summer schools

Linguistic Society of America (LSA): The NAACL sponsored 5 scholarships, and the ACL contributed as well. The applicants included 3 NAACL students and 50 from LSA students. Potential reasons for this imbalance: summer schools aren't part of the NAACL community's tradition; the courses might have seemed too introductory for students within the NAACL community; lack of advertising or poorly timed advertising among the NAACL community. On the other hand, the great response from the LSA side of things was quite welcome. It would be good to determine some way to streamline the application process and the decision process, in a way that works with the procedures the LSA already has in place.

Johns Hopkins summer school: The decision from last year was to allocate $5K for this year, amounting to about $1K per student. Last year, different amounts of support were allocated to different students on a (perceived) need basis. There was some question of whether we couldn't have the allocation depend on the number of worthy applicants, or specify a more loose target, such as "up to ten". A committee was created to handle the JHU scholarship program, consisting of Bill Dolan, Mary Harper, and Owen Rambow.

(return to top of minutes)

[8] Miscellaneous topics

There may be some perception that NAACL is more applied than the ACL; this has been expressed in reviews, for instance ("this is an ACL paper, not an NAACL paper").

It was noted that workshops bring in money because they tend to result in larger attendance. Perhaps some workshops could be oriented more towards outreach towards other areas.

(return to top of minutes)



Date: April 22, 2007
Place: Hyatt Regency, Rochester
Attendees: Jennifer Chu-Carroll, Bill Dolan, Mary Harper, Graeme Hirst, Lillian Lee, Christopher Manning, Owen Rambow
Guests: HLT-NAACL 2007 General Chair Candace Sidner and NLP Program Chair Matthew Stone, and Business Manager Priscilla Rasmussen