The Third Workshop on Figurative Language Processing

Shared Tasks

The workshop will host two shared tasks on grounded understanding of figurative language and on euphemism detection.

A Shared Task on Understanding Figurative Language

In recent years, there have been several benchmarks dedicated to figurative language understanding, which generally frame ``understanding'' as a recognizing textual entailment (a.k.a natural language inference (NLI)) task --- deciding whether one sentence (premise) entails/contradicts another (hypothesis).

For this shared task, our task prediction requires that the model not only correctly infer the whether a given premise entails or contradicts the hypothesis, but also explain why using natural language explanations that are comprehensible to an end-user assessing model's reliability. Our task covers understanding of a variety of figurative language such as sarcasm, simile, metaphor, and idiom.

For more information about the shared task visit https://figlang2022sharedtask.github.io/task.html.

We are hosting the shared task on CodaLab. For participation go here.

Important dates:

  • July 10, 2022: CodaLab competition is open; training data and auxiliary scripts can be downloaded

  • Aug 15, 2022: Test data can be downloaded and results submitted; performance will be tracked on CodaLab dashboard

  • Aug 20, 2022: Last day for submitting predictions on test data

  • Sept 7 14, 2022: Papers describing the systems are due

  • Oct 9 12, 2022: Notification of acceptance

  • Oct 23, 2022: Camera-ready papers due

  • December 8, 2022: Workshop

Contact:


Euphemism Detection shared task

Euphemisms are mild or indirect expressions used in place of harsher or more offensive ones. Euphemisms are often used to mask profanity or refer to taboo topics such as death, disability, sex, religion or personal relationships in a polite way. Euphemisms are often ambiguous: their literal and non-literal interpretation is context-dependent:


Asked to choose between jobs and the environment, a majority -- at least in our warped, first-past-the-post system -- will pick jobs. [non-euphemistic]

vs.

This summer, the budding talent agent was between jobs and free to babysit pretty much any time. [euphemistic]


The state of the art language models perform well on many major NLP benchmarks; however, it is unclear how such models perform on euphemisms. Thus, we propose a euphemism detection task: given an input sentence, identify whether the sentence contains a euphemism.


For more information about the shared task and to participate visit https://codalab.lisn.upsaclay.fr/competitions/5726.

Important dates:

  • July 5, 2022: CodaLab competition is open; training data and auxiliary scripts can be downloaded

  • Aug 5, 2022: Test data can be downloaded and results submitted; performance will be tracked on CodaLab dashboard

  • Aug 20, 2022: Last day for submitting predictions on test data

  • Sept 7, 2022: Papers describing the systems are due

  • Oct 9, 2022: Notification of acceptance

  • TBD, 2022: Camera-ready papers due

  • December 8, 2022: Workshop

Contact: