Thank you, Tristan!
Looking at the submission form, it seems like we need a "short summary." The 2027 template
says abstract is not required so we didn't include one - but now we need to write one for the form. Here's what I suggest (75% is adapted from last year's):
Modeling humor, deemed “AI-complete” due to its reliance on context, cultural knowledge, and inference, remains a challenge—and a potential test case for AGI—despite generative AI advances. The 3rd Workshop on Computational Humor (CHum 2027) fosters work
on modeling humor with current methods in computational linguistics and NLP, against the backdrop of humor theory and with reference to textual, visual, and multimodal corpora. A central open question is whether LLMs understand humor in a human-like way (internal
representations of setup/punchline, incongruity, script opposition, and cultural inference) or mainly match surface patterns. Evidence is mixed on explanation, detection, rating, and generation. We particularly welcome theory-driven probes, multimodal work,
and applications such as joke generation and humor-preserving computer-assisted translation. The full-day workshop emphasizes diverse, multilingual, and interdisciplinary perspectives.
I am not attached to it, please edit whatever you don't like/can improve. If
everyone's happy with
it and the proposal, I'll submit.
Ori