2022Q3 Reports: Tutorial Chairs

Summary

This year, as has been the tradition over the past few years, the call, submission, reviewing, and selection of tutorials were coordinated jointly for multiple conferences: ACL, NAACL, COLING, and EMNLP.

• ACL: Luciana Benotti (until Jan 2022 when she resigned to join the NAACL board), Naoaki Okazaki, Yves Scherrer (after Jan 2022), Marcos Zampieri
• NAACL: Cecilia O. Alm, Yulia Tsetkov, Miguel Ballesteros
• COLING: Heng Ji, Hsin-Hsi Chen, and Lucia Donatelli
• EMNLP: Samhaa R. El-Beltagy and Xipeng Qiu

The call for proposals was based on the one used in 2021. The major changes are improved descriptions of diversity and inclusion (D&I) and the page limit (4 pages of content plus unlimited pages for references) of a tutorial proposal. We decided to use OpenReview instead of START as a submission management system, following the decision for paper submissions for the main conference of ACL 2022.

We formed a review committee of 34 members, including the tutorial chairs of the conferences and 23 external reviewers (see Program Committee of the Tutorial Abstracts volume of the proceedings for the full list). A reviewing process was organized so that each proposal received three reviews. The selection criteria included clarity and preparedness, novelty or timely character of the topic, instructors' experience, likely audience interest, open access of the tutorial instructional material, and diversity and inclusion. A total of 47 tutorial submissions were received, of which eight were selected for presentation at ACL.

• T1: A Gentle Introduction to Deep Nets and Opportunities for the Future [introductory]
• T2: Towards Reproducible Machine Learning Research in Natural Language Processing [introductory]
• T3: Knowledge-Augmented Methods for Natural Language Processing [cutting-edge]
• T4: Non-Autoregressive Sequence Generation [cutting-edge]
• T5: Learning with Limited Text Data [cutting-edge]
• T6: Zero- and Few-Shot NLP with Pretrained Language Models [cutting-edge]
• T7: Vision-Language Pretraining: Current Trends and the Future [cutting-edge]
• T8: Natural Language Processing for Multilingual Task-Oriented Dialogue [cutting-edge]

We used aclpub2 for generating the proceedings of the Tutorial Abstracts volume and the conference handbook. Because this was the first year of using aclpub2, we received kind support from the publication chair Danilo Croce and the handbook chair Marco Polignano. The ecosystem of aclpub2 was convenient and useful, but a tutorial chair may need a bit of knowledge about Python (especially for package installation), LaTeX, and YAML.

Initially, we expected that tutorial instructors would give presentations live at the conference venue. However, the spread of the omicron variant of COVID-19 confounded the expectation. We decided to allow tutorial presentations either in on-site, hybrid, or virtual mode. As with other chairs, we participated in meetings with Underline to prepare the facility for both on-site and remote presentations. This also affected the deadline for preparing presentation slides because Underline requested presentation slides earlier for preparing the Web site of the virtual conference.

Lessons learned

• We should have rehearsed the process of proposal submission, reviewer assignment, making decisions, and sending out notifications on OpenReview before opening the submission site. OpenReview was not so friendly to program chairs. It was difficult to test the functionality after we started the reviewing process.
• It may be worth reconsidering the joint call from conferences for the whole year. We would suggest calling for tutorial proposals twice per year. Collecting and reviewing tutorial proposals at one time per year is efficient (in terms of reviewing effort and reducing the number of rejected proposals), but,
  • The details of later conferences (e.g., chairs, venue, etc.) may not be fixed.
  • If a tutorial proposal was rejected from the joint reviewing process, the tutorial instructors would lose the opportunity in *CL conferences for a whole year.
  • The research trend changes rapidly (new exciting research may appear within a year).
• The experience from past tutorial chairs is valuable in this process as it takes time for new tutorial chairs to understand the process. SemEval has a committee model in which at least one chair stays in the committee as a senior chair in the subsequent year to pass along information to the new chairs. A similar model could be discussed for ACL tutorials and workshops.