mlsanityData to 7 Oct 2026

BottleHumor: Self-Informed Humor Explanation using the Information Bottleneck Principle

arXiv 2502.18331 ↗First posted 22 Feb 2025cs.CLcited by 6

Not on the map. Why?

Abstract (arXiv)

Humor is prevalent in online communications and it often relies on more than one modality (e.g., cartoons and memes). Interpreting humor in multimodal settings requires drawing on diverse types of knowledge, including metaphorical, sociocultural, and commonsense knowledge. However, identifying the most useful knowledge remains an open question. We introduce method, a method inspired by the information bottleneck principle that elicits relevant world knowledge from vision and language models which is iteratively refined for generating an explanation of the humor in an unsupervised manner. Our experiments on three datasets confirm the advantage of our method over a range of baselines. Our method can further be adapted in the future for additional tasks that can benefit from eliciting and conditioning on relevant world knowledge and open new research avenues in this direction.

Key facts. Cited by 6 papers in the map (data through 7 Oct 2026). Builds on 29 of its 57 references that are in the map. First posted on arXiv 22 Feb 2025.

Builds on

29 of 57 references.

References · 29

  1. 2023Self-Refine: Iterative Refinement with Self-Feedbackcited by 3,317
  2. 2024Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phonecited by 1,883
  3. 2022Learn to Explain: Multimodal Reasoning via Thought Chains for Science Question Answeringcited by 1,881
  4. 2024Justice or Prejudice? Quantifying Biases in LLM-as-a-Judgecited by 354
  5. 2024Visual Sketchpad: Sketching as a Visual Chain of Thought for Multimodal Language Modelscited by 260
  6. 2019Towards Multimodal Sarcasm Detection (An _Obviously_ Perfect Paper)cited by 173
  7. 2020Graph Information Bottleneckcited by 168
  8. 2020Unsupervised Commonsense Question Answering with Self-Talkcited by 161
  9. 2019UR-FUNNY: A Multimodal Language Dataset for Understanding Humorcited by 130
  10. 2021Reframing Human-AI Collaboration for Generating Free-Text Explanationscited by 125
  11. 2022Do Androids Laugh at Electric Sheep? Humor "Understanding" Benchmarks from The New Yorker Caption Contestcited by 98
  12. 2023How Does Information Bottleneck Help Deep Learning?cited by 92
17 more
  1. 2024KAM-CoT: Knowledge Augmented Multimodal Chain-of-Thoughts Reasoningcited by 60
  2. 2022Multimodal Information Bottleneck: Learning Minimal Sufficient Unimodal and Multimodal Representationscited by 48
  3. 2023MemeCap: A Dataset for Captioning and Interpreting Memescited by 46
  4. 2024Molecular Facts: Desiderata for Decontextualization in LLM Fact Verificationcited by 42
  5. 2021Compressive Visual Representationscited by 41
  6. 2024TokenSHAP: Interpreting Large Language Models with Monte Carlo Shapley Value Estimationcited by 41
  7. 2019BottleSum: Unsupervised and Self-supervised Sentence Summarization using the Information Bottleneck Principlecited by 39
  8. 2023Reverse Engineering Self-Supervised Learningcited by 33
  9. 2023SMILE: Multimodal Dataset for Understanding Laughter in Video with Language Modelscited by 28
  10. 2024Deductive Closure Training of Language Models for Coherence, Accuracy, and Updatabilitycited by 22
  11. 2024Understanding Figurative Meaning through Explainable Visual Entailmentcited by 19
  12. 2023Natural Language Decompositions of Implicit Content Enable Better Text Representationscited by 18
  13. 2015We Are Humor Beings: Understanding and Predicting Visual Humorcited by 15
  14. 2023Dynamic Multimodal Information Bottleneck for Multimodality Classificationcited by 14
  15. 2021Leveraging Information Bottleneck for Scientific Document Summarizationcited by 13
  16. 2024Cracking the Code of Juxtaposition: Can AI Models Understand the Humorous Contradictionscited by 11
  17. 2024Locating Information Gaps and Narrative Inconsistencies Across Languages: A Case Study of LGBT People Portrayals on Wikipediacited by 10

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 2; 2026: 4.

6 papers.

  1. 2025Large Language Models for Subjective Language Understanding: A Surveycited by 9
  2. 2025SWI: Speaking with Intent in Large Language Modelscited by 0
  3. 2026Culture-Aware Humorous Captioning: Multimodal Humor Generation across Cultural Contextscited by 0
  4. 2026Grounded Satirical Generation with RAGcited by 0
  5. 2026Computational Humor with Multimodal LLMs: Methods, Datasets, Evaluation, and Challengescited by 0
  6. 2026Towards High-Level Semantic Intelligencecited by 0