Student Sessions at ACL Conferences

The Student Session is an established feature of ACL conferences. However, there is no accepted format for this part of the conference. This report outlines a new policy for the operation of the student sessions, the aim being to:


Goal of the Student Sessions
The Student Session provides a valuable opportunity for the next generation of computational linguists to enter the CL community. It allows the best students in the field to take their first important step towards becoming professional computational linguists by:

For the ACL Student Sessions to achieve this goal consistently, we need to ensure that we attract papers from the best students internationally, provide professional reviews that (a) help the students to develop their work and (b) assist the programme committee in selecting the best papers. We also need to provide a suitable venue for their presentation. Finally, we need to ensure that the student session is very well attended by the participants of the main conference. Over the years the student session has taken a variety of forms: as sessions like any other within the main conference, as special lunchtime sessions, as poster sessions during breaks, and as a 1-day pre-conference workshop. They have not been equally sucessful. The chosen format has, to a large extent, been guided by the number of submissions to the Student Session, but also by the number of paper accepted in the main conference and thus the number of session slots available for student presentations. Funding considerations also have a role to play; for example, at ANLP/NAACL'2000 a substantial amount of funding came from the NSF, but this brought with it the condition that the student session be run as a Workshop separate from the main conference.

The following table shows the available statistics from 1991:
 

Year Submitted Accepted Format
1991 unknown 14 4 sessions, 2 at a time, 20 mins per student
1992 48 20 6 sessions, 2 at a time, 20 mins per student
1993 30 11 1 session, 18 minutes per student 
1994 41 10 2 sessions in parallel, 20 minutes per student 
1995 48 19 poster session, 2 hours and 30 minutes \
1996 32 14 poster session, 1 hour and 45 minutes 
1997 42 10 2 sessions, 20 minutes per student \
1998 46 12 4 lunch-time sessions, 20 minutes per student \
1999 30 10 4 sessions, 2 at a time, 25 minutes per student \
2000(ANLP/NAACL) 18 8 pre-conference workshop, 25 min presentation, 15 min discussion led by 2 senior researchers
2000 36 10 4 sessions, 2 at a time, 18 min presentation, 12 min discussion led by 2 senior researchers 

While recognising the need to build some flexibility into the process, the picture above makes it difficult for outsiders to judge the status of a paper in an ACL student session. In general, papers presented in sessions that are interleaved with the main conference sessions, or in a dedicated pre-conference Workshop, are viewed more favourably than poster presentations. A related problem arises with the form of the publication. Student papers and posters generally appear within the main conference proceedings. These are often indistinguishable from papers in the main sessions, and even in cases where they are not (e.g., identified by a different page numbering system), there is a general tendency for authors and others to cite them as though they were regular ACL papers (The same is true of the extended abstracts for the Demo Session). This practice has the potential to lower the perceived quality of ACL papers. The attendance at the student session tends to be unaffected by the format of the presentations. However, the quality of the feedback to the students tends to be much lower for poster sessions.

ACL Policy for Student Sessions
Format
The organising committee will be co-chaired by two doctoral students in Computational Linguistics, and will be coordinated by a Faculty Advisor. The members of the organising committee will be appointed by the General Chair, who will consult the co-chairs of the most recent Student Workshop for recommendations on the new student co-chairs.

The process of reviewing and selecting papers will conform to the standard ACL policy on programme committees, with the exception that each paper should be reviewed by a combination of students and established figures in the field.

In order to maintain a consistent status, the student session, will be run as either a workshop alongside the other workshops of the conference, or as a special session interleaved within the main conference. The final choice of format from among these for any given meeting will be determined by the programme chair of the main conference in consultation with the co-chairs of the student session, based on the number of available slots in the main conference, the number of accepted papers for the student session, and any other constraining considerations (e.g., the requirements of external funders).

The recent innovation of appointing panelists for each paper is a good idea. At the very least, it guarantees a good discussion on each paper. We recommend that this be continued, although the amount of time allocated to each panelist will obviously be much more limited when the session is interleaved with the main conference.

Scheduling
If the session is run with the conference workshops, the length of slots for each presenter can be anywhere from 25 -- 45 minutes, depending on the preferred format of the organisers. However, if it is interleaved with the main sessions of the conference, then the length of presentations slots will be no longer than those of the main conference.

Title
The title of the session will be Student Research Workshop. As an indicator of a separate existence, but closely linked to the main conference, this title clearly distinguishes between papers accepted for the main conference and the student session, and provides more opportunities for external funding (e.g., from NSF and the European Commission).

Publication
At the meeting of the ACL executive committe in May, 2000 in Seattle, it was decided that future proceedings will be composed of two parts:

These two parts will be distributed as an integral set of volumes, both at the conference itself and for any post-conference sales.

This policy has already been put into effect for the ACL'2000 meeting. The title of the companion volume will replace ``Proceedings of the Conference'' with:

Companion Volume to the Proceedings of the Conference
including the
Proceedings of the Student Research Workshop
Tutorial Abstracts and Demonstration Notes

Qualifications
The student session will be dedicated to the presentation of graduate work in progress or significant undergraduate research. It will therefore normally be open only to 1) graduate students who have settled on their thesis direction but who still have significant research left to do, and 2) undergraduate students who have pursued an undergraduate research project.  Those graduate students in the final stages of their thesis should be submitting instead to the main conference. To enforce this, submissions must be accompanied by a curriculum vitae and/or a letter from the thesis advisor confirming the current state of the thesis and giving an estimated date of submission.

A student who has already presented at an ACL/EACL/NAACL student session will not be allowed to present again at the student session of any of these conferences, but encouraged to submit instead to the main conference.

Roles and Responsibilites of the Organising Committee
What follows is a non-exhaustive list outlining the main responsibilites of the organising committee.

The Faculty Advisor:


The co-chairs:

Approved by ACL Exec: October 2000