Call For Papers

UM-IoS Workshop 2022

The first Unimodal and Multimodal Induction of Linguistic Structures Workshop (UM-IoS) will be held in conjunction with EMNLP 2022. The one-day workshop will occur during EMNLP’s workshop period (December 7th or 8th, 2022). UM-IoS provides a discussion platform on induction of structures (IoS), bringing together researchers and practitioners across natural language processing, computer vision, machine learning and general AI fields. The full-day event includes invited talks, oral presentations, and poster sessions.

Workshop Description

Induction of structures (IoS) is the process of inducing structured objects (a general term of structured data rather than discrete or real values) from a set of observations. It is a branch of machine learning where the output space consists of discrete combinatorial objects (such as strings, trees, and graphs) and is unobserved or partially observed during learning. IoS in natural language processing has often been very focused on the problem of uncovering the syntactic structure (e.g., a constituent or dependency tree), semantic structure, label sequence, discourse structure etc from input text. Such structures have been found useful in downstream tasks such as information extraction and machine translation. Apart from the wide usage in language processing, inducing the underlying structures from raw sensory inputs (e.g., vision) has been a long-standing challenge in the field of artificial intelligence.


Relevant topics for the workshop include, but are not limited to, the following areas:

  • Understanding and discovering the latent structures (e.g., sequence, tree, graph), including linguistic or non-linguistic structures from various text and sensory data.
  • Probing and analysis of structures from various models, for example foundation models.
  • New structured learning approaches. New methodologies including both statistical approaches and neural approaches, ranging from semi-supervised, weakly-supervised setting, zero-shot and few-shot learning, transfer learning, low-resource transfer, unsupervised adaptation, to purely unsupervised approaches.
  • Applications of induced structures to different downstream tasks in different AI fields (e.g., language, vision, robotics, cognition), as well as for computational modeling in language acquisition and psycholinguistics.
  • Investigating and understanding joint structure modeling of multiple modalities. For example, fusing other modalities’ information (e.g., visual information) to enhance linguistic structures or vice versa.
  • New datasets, benchmarks or libraries for implementations and evaluation methods of IoS.

Submission guidelines

Long/short papers should consist of eight/four pages of content plus unlimited pages for bibliography. Submissions must be in PDF format following the official ACL style templates, anonymized for review. All long and short papers must follow the ACL Author Guidelines. Paper submissions must use the official ACL style templates, which are available from here (Latex and Word). Please follow the paper formatting guidelines general to “*ACL” conferences available here. Each submission will be reviewed by at least 2 program committee members.

Submission is electronic and is managed by two ways: OpenReview at and Softconf system at A paper may not be simultaneously under review through ARR and Softconf.

Important Dates (anywhere on Earth/UTC-12)

We accept submissions from both ARR and Softconf. So if you find that you have missed the ARR deadline, feel free to directly submit it to Softconf.

ARR Submission deadlineJuly 15, 2022
ARR Commitment deadlineOctober 2, 2022
Softconf Submission deadlineOctober 2, 2022(*UPDATED*)
Acceptance notificationOctober 20, 2022 (TBD)
Workshop DateDecember 7th/8th, 2022 (TBD)

Contact information


E-mail: UM-IoS[at]

Read more

[3] - EMNLP 2022