December 21, 2020 - First Call for Papers
February 15, 2021 - Second Call for Papers
March 26, 2021 - Anonymity period begins
May 7, 2021 - *Extended* Submission Deadline
May 28, 2021 - Notification of Acceptance
June 7, 2021 - Camera-ready papers due
August 5, 2021 - Workshop Day
All deadlines are 11.59 pm UTC-12 (anywhere on Earth)
As the impact of machine learning on all aspects of our lives continues to grow, the need for systems that learn through interaction with users and the world becomes more and more pressing. Unfortunately, much of the recent success of NLP relies on large datasets and extensive compute resources
to train and fine-tune models, which then remain fixed. This leaves a research gap for systems that adapt to the changing needs of individual users or allow users to continually correct errors as they emerge. Learning from user interaction is crucial for tasks that require a high grade of personalization and for rapidly changing or complex, multi-step tasks where collecting and annotating large datasets is not feasible, but an informed user can provide guidance.
We define Interactive Learning for NLP as training, fine-tuning or otherwise adapting an NLP model to inputs from a human user or teacher. Relevant approaches range from active learning with a human in the loop, to training with implicit user feedback (e.g. clicks), dialogue systems that adapt to user utterances, and training with new forms of human input. Interactive learning is the converse of passive learning from datasets collected offline with no human input during the training process.
We encourage submissions in the following topics, including but not limited to:
Interactive machine learning methods, theory and practice: from active learning with a user to methods that extract, interpret and aggregate user feedback or preferences from complex interactions, such as natural language instructions.
User effort: the amount of user effort required for different types of feedback and how users cope with the system misinterpreting instructions.
Different kinds of user feedback: beyond providing training labels, users may influence and interrogate NLP models in different ways, such as providing natural language instructions or implicit feedback through mouse clicks.
Evaluation methods: approaches to assessing interactive methods, such as low-effort, easily reproducible methods with real-world users and simulated user models for automated evaluation.
Reproducibility: procedures for documenting user evaluations and ensuring they are reproducible.
Data: novel datasets for training and evaluating interactive models.
Empirical results that investigate scenarios where interactive learning is effective.
Submission is electronic, using the Softconf START conference management system:
https://www.softconf.com/acl2021/w09_internlp2021/
Please note that we have now fixed an issue with the submission form that prevented you from submitting.
Both long and short papers must follow the ACL 2021 style file: https://2021.aclweb.org/calls/papers/#paper-submission-and-templates
We solicit three categories of papers:
Standard workshop papers: anonymized submissions describing substantially original research not previously published in other venues.
Extended abstracts: anonymized submissions describing preliminary but interesting ideas or results not previously published in other venues.
Cross-submissions: non-anonymized papers on relevant topics that have previously been accepted for publication in another venue.
Only standard workshop papers will be included in the proceedings as archival publications. Please note, that standard workshop papers must follow the ACL submission policy. All three categories of papers may be long (maximum 8 pages plus references) or short (maximum 4 pages plus references).
Each submission can be accompanied by a single PDF appendix. The paper submissions need to remain fully self-contained, as these supplementary materials are completely optional, and reviewers are not even asked to review or download them. Supplementary materials need to be fully anonymized to preserve the double-blind reviewing policy.
Dual Submission with ACL is allowed. You will be asked to specify the cross-submission information in the START system. If the paper is accepted to both ACL and the workshop, it will be considered as a cross-submission and will not be a part of the workshop proceedings. If it gets rejected from ACL and accepted to the workshop and you want us to consider it as a standard workshop paper which will be included in the proceedings, please contact us to let us know.
Yoav Artzi, Cornell University and ASAPP, Inc.
Dan Goldwasser, Purdue University
Percy Liang, Stanford University
Dan Roth, University of Pennsylvania
Dorsa Sadigh, Stanford University
Tania Bedrax-Weiss, Google
Ido Dagan, Bar-Ilan University
Dilek Hakkani-Tur, Amazon Alexa AI
Seung-won Hwang, Yonsei University
Julia Kreutzer, Google Research
Alison Smith-Renner, University of Maryland
Yu Zhou, University of California, Davis
Kianté Brantley, University of Maryland College Park
Soham Dan, University of Pennsylvania
Iryna Gurevych, Technische Universität Darmstadt
Ji-Ung Lee, Technische Universität Darmstadt
Filip Radlinski, Google
Hinrich Schütze, Ludwig Maximilians Universität Munich
Edwin Simpson, University of Bristol
Lili Yu, ASAPP, Inc.