SRW tasks

Summary

This page summarizes all tasks that have to be carried out during the organization of the ACL Student Research Workshop. Please refer to the SRW page for more information.

SRW tasks

Call for Papers and publicity

The SRW must be made public very early, in order to encourage students to submit their work, and in order to encourage professors to encourage their students to submit their work. This should be done not only by submitting the CFP to lists, but also by sending personal messages to faculties, in order to increase the awareness of the SRW and its benefits. Don't forget to mention the (partial!) student travel funding as one of the major benefits of the workshop, as well as the personal tutoring by senior researchers. Ask PC members to link the workshop on their homepages/faculty pages. Before you post the call, the general conference chair should see it.

Website and START

In the last years, the SRW website was located within the main conference website and administered by the main conference site webmaster. It should contain the call for papers, including the members of the program committee, as well as the panelists and the accepted papers (as soon as fixed). We also added an NSF logo to acknowledge their sponsorship.

The SRW gets its own START site. The START system can be used to manage the entire reviewing process. Setting up this page as early as possible and getting familiar with the system is beneficial. START is provided to the SRW by the ACL. James Clarke (U Edinburgh) in place of Simone Teufel (U Cambridge) set it up for us.

Budget and NSF grant

Traditionally, the SRW gets funded by the National Science Foundation. The grant is used for funding of the travel of the participants. Usually, the faculty advisor solicits the fund at the NSF, the earlier, the better; the faculty advisor's university will be the holder of the fund.

The funding proposal was submitted to NSF before the paper submission deadline. Therefore we estimated the amount of the proposed funding using the previous year's statistics, i.e. 15 participants (5 from North America, 5 from Europe, 2 from Asia, ...). The airfares to Columbus, OH was calculated for these participants based on projected airfares from popular internet travel sites. We calculated the accommodation cost of each participant for a 4 nights stay in a double shared room in a hotel near the conference venue. The registration costs were also estimated from 2007. We also estimated some amount to contribute to ACL for the organizational expenses such as room rents, rent of the technical equipments, etc.

In 2008 we had less participants than we had expected and more participants from North America. Therefore, we could also allocate an additional $100-150 for each participant for meals and additional expenses. In order to distribute the funding fairly among the participants, we sent e-mails to the participants and asked their expected costs for the airfares. Then personal e-mails were sent to the participants to let them know the upper limit of their grants.

We asked the students to send us a breakdown of their costs after the conference. The amount for each participant was reallocated before the reimbursements. This year we were able to reimburse all participants for all of their costs since we overestimated the NSF funding (especially travel costs) using 2007's statistics. Note that possibly only partial reimbursements can be made if there are many participants or many participants have a long way to travel.

Coordinating the reimbursement of the students

In 2008, Priscilla Rasmussen kindly offered to handle the reimbursements to students. Then, she will bill University of Pittsburgh (holder of the NSF grant) for one amount. Instructions should be given to the participants about what information to send where to get their reimbursements. The ACL and the SRW will not pay for funds to be disbursed in non-US funds. If someone requests that, the cost of conversion will need to come out of that participants' allotted amount.

Program committee and reviewing process

We recruited the program committee out of members of previous year's program committees, known senior researchers, personal contacts and recommended people. A program committee member is not supposed to serve for more than two times. We did not have to let the ACL exec authorize the PC (even though the ACL handbook says differently).

The program committee should be composed of students and senior researchers. We had 61 reviewers (30 students, 31 senior researchers) and 27 submissions, which allowed us to assign five reviewers to a single paper (three senior researchers and two students). This meant at most three papers to review for each senior PC member and two papers to review for each student member. The high number of the reviews was highly praised by the authors after the workshop. We tried to select the reviewers such that all areas were sufficiently covered. We used a list of keywords to match the submissions to the reviewers. This was difficult and a lot of work. Especially if you have more submissions, consider using the bidding facilities in START, which allows the reviewers to express their preferences before you make the final distribution. START also offers you some other choices on the review process (e.g., control the degree of visibility of reviewers within the PC). Remember that you have to decide on all details of the reviewing process before the submission deadline. Since many of the reviewers will also review papers for the ACL main conference and the SRW has its own START page, also don't forget to mention that the SRW papers will not appear in the main conference START reviewer accounts.

Proceedings

The co-chairs are responsible for the compilation of the proceedings and the CD-ROM/ACL anthology version of the SRW proceedings. ACL offers great software, called ACLPUB, which automates parts of that process. However, preparing the book is a big chunk of work for which enough time should be allocated, mainly because authors tend not to follow formatting instructions. We could find all sorts of errors in the camera-ready versions of the papers, ranging from small ones, such as overful columns, to severe font problems. Since you have to anticipate two or even more rounds of revisions, set the deadline for the SRW authors for the submission of their camera-ready version early enough, at least 2 weeks before the publication chair deadline for the delivery of the finished proceedings.

Panelists

It is a tradition at the SRW to ask senior research to act as panelists for the SRW talks. In past years, one or two panelists were assigned to each paper. A panelist has to read the student's paper and prepare in-depth questions which should allow the student to advance in his research. A separate 5-min question session at the end of each talk is reserved to panelists questions, before the general public can ask its questions.

In past years, a general invitation along with the list of abstracts of the student papers was sent to a long list of senior researchers. Panelists posts were then distributed on a first-come-first-serve basis. We, in contrast, tried to search researchers which matched very well with a specific paper topic and asked them directly to be the panelist for a single paper. Both methods seem to work, although the first method is maybe a more feasible option if you have many submissions. We think that responses are more likely if in the invitation, it is made clear that that being a panelist is not as much work as being a reviewer.

Schedule

In the last years, efforts have been made to make the SRW look like a part of the main conference, not like a separate workshop. The oral session(s) have been scheduled in parallel to main conference sessions and SRW posters were hung indistinguishably along with main conference posters. This has proven to be beneficial for the SRW presenters, due to an increased amount of feedback. The program chair(s) should be informed about this.

Coordination with conference organizers and local arrangements

1. coordinate with the program co-chairs to schedule the talks and posters

2. coordinate with the general co-chairs about costs the SRW should pay for room, equipment, food, etc. Typically, the SRW contributes some of the funds from its NSF grant to cover these costs, but it seems this is handled on a volunteer basis.

3. coordinate with local arrangements about room setup for the SRW spoken sessions, including where the panelists should sit.

4. coordinate with local arrangements about setup for the poster session.

5. make it clear to the general co-chair and local arrangements that the student lunch ACL sponsors is completely separate from the SRW. The SRW does not pay for it or coordinate it. The SRW may have its own lunch, but that is a different lunch. This was a source of confusion for everyone in 2008.

Miscellaneous

Apart from the traditional ACL Student Lunch, we gathered with all participants, organizers and panelists of the SRW for a separate lunch on separate day. This was a great occasion to get to know each other more personally, and at the same time easy to organize.

SRW papers and presenters

1. A Supervised Learning Approach to Automatic Synonym Identification based on Distributional Features, by Masato Hagiwara

2. An Integraged Architecture for Generating Parenthetical Constructions, by Eva Banik

3. Impact of Initiative on Collaborative Problem Solving, by Cynthia Kersey

4. A Subcategorization Acquisition System for French Verbs, by Cedric Messiant

5. Arabic Language Modeling with Finite State Transducers, by Ilana Heintz

6. The role of positive feedback in Intelligent Tutoring Systems, by Davide Fossati

7. Inferring Activity Time in News through Event Modeling, by Vladimir Eidelman

8. Combining Source and Target Language Information for Name Tagging of Machine Translation Output, by Shasha Liao

9. An Unsupervised Vector Approach to Biomedical Term Disambiguation: Integrating UMLS and Medline, by Bridget McInnes

10. Adaptive Language Modeling for Word Prediction, by Keith Trnka

11. A Hierarchical Approach to Encoding Medical Concepts for Clinical Notes, by Yitao Zhang

12. A Re-examination on Features in Regression Based Approach to Auto-matic MT Evaluation, by Shuqi Sun, Yin Chen and Jufeng Li