---
title: "‘Who would have thought of that!' — paper by Joshi et al. (2016)"
description: "Joshi et al. treat sarcasm detection as a topic-modelling problem, which they believe is the first time this… 23 references, cited by 1 paper here."
authors: ["Aditya Joshi", "Prayas Jain", "Pushpak Bhattacharyya", "Mark Carman"]
year: 2016
venue: "Proceedings of the Workshop on Extra-Propositional Aspects of Meaning in Computational Linguistics (ExProM)"
source: "https://aclanthology.org/W16-5001/"
arxiv: "1611.04326"
licence: "CC BY 4.0 (ACL Anthology)"
licence_url: "https://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/joshi-2016-who-would-have-thought-that/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/joshi-2016-who-would-have-thought-that/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2016](https://md.research.elamur.ai/free/en/papers/2016/page/4/) / ‘Who would have thought of that!': A Hierarchical Topic Model for Extraction of Sarcasm-prevalent Topics and Sarcasm Detection

# ‘Who would have thought of that!': A Hierarchical Topic Model for Extraction of Sarcasm-prevalent Topics and Sarcasm Detection

**Authors:** [Aditya Joshi](https://md.research.elamur.ai/free/en/authors/aditya-joshi/), Prayas Jain, [Pushpak Bhattacharyya](https://md.research.elamur.ai/free/en/authors/pushpak-bhattacharyya/) and [Mark Carman](https://md.research.elamur.ai/free/en/authors/mark-carman/) · **Year:** 2016 · **Venue:** [Proceedings of the Workshop on Extra-Propositional Aspects of Meaning in Computational Linguistics (ExProM)](https://md.research.elamur.ai/free/en/venues/workshop-on-extra-propositional-aspects-of-meaning-in-computational-linguistics/) ([Workshop on Extra-Propositional Aspects of Meaning in Computational Linguistics 2016](https://md.research.elamur.ai/free/en/venues/workshop-on-extra-propositional-aspects-of-meaning-in-computational-linguistics/2016/)) · **Publisher:** [International Committee on Computational Linguistics (ICCL)](https://md.research.elamur.ai/free/en/publishers/international-committee-on-computational-linguistics-iccl/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/W16-5001/) · [PDF](https://aclanthology.org/W16-5001.pdf) · [arXiv 1611.04326](https://arxiv.org/abs/1611.04326) · [OpenAlex](https://openalex.org/W2556753300)

**Citations:** 4 citations (OpenAlex, 2026-10-14) · [1](https://md.research.elamur.ai/free/en/papers/joshi-2016-who-would-have-thought-that/cited-by/) in our library, of the 141,840 papers whose bibliographies we read

## Summary
Joshi et al. treat sarcasm detection as a topic-modelling problem, which they believe is the first time this has been done. Their starting intuition has two parts. Some subjects attract sarcasm far more than others. And sarcastic tweets tend to mix positive and negative words, whereas literal tweets mostly stay with one polarity.

The model is a hierarchical, supervised variant of LDA, trained with Gibbs sampling on 166,955 tweets. The tweets carry hashtag-derived labels: literal positive, literal negative, or sarcastic. Each tweet gets one latent topic. Each word gets a latent switch that determines whether it belongs to the topic vocabulary or carries sentiment, and sentiment words also get a latent polarity.

The learned distributions point to work, weather and gun laws as sarcasm-heavy topics. The polarity mixture the model estimates differs clearly across the three labels. Used as a classifier, the model reaches about 46–47% F, against roughly 19% for two feature-based baselines. This suggests that thematic context plus sentiment mixture is a useful signal for spotting irony.

## Key points
- **Data:** 166,955 tweets, labelled through hashtags into three classes (positive, negative, sarcastic). The model assumes one topic per tweet and uses S = 2 word polarities.
- **Model structure:**
  - Sentiment and word distributions are chained hierarchically, for example a topic-specific word distribution drawn around a general sentiment-specific one.
  - The topic/sentiment switch prior η_w is estimated in advance from a large corpus and is not resampled.
- **Topics by label probability p(l|z):**
  - Holidays (0.9538) and Father's Day (0.9224) are strongly positive.
  - The Orlando shooting topic is about 0.95 negative.
  - Gun laws are the most sarcasm-prevalent topic shown (0.5230), with work and humour also sarcasm-heavy.
- **Sentiment mixture:**
  - Over 60% of negative tweets contain no words sampled as positive.
  - Around 20% of sarcastic tweets have roughly half positive words.
- **Detection results (F-score):**
  - Sampling-based method: 46.80 (P 45.94, R 47.70).
  - Log-likelihood method: 46.48.
  - Buschmeier et al. (2014): 18.85.
  - Liebrecht et al. (2013): 19.86.
  - Both baselines have recall close to 100% but precision around 10–11%.

## Interesting ideas
- A word's tendency to be evaluative rather than content-bearing is learned once from a big corpus and then held fixed, which keeps the sampler from wandering and makes the topic/sentiment split stable.
- Polarity clash inside one short text (a positive evaluation applied to a negative situation, as in "love being ignored") is used as the core irony cue. Hyperbole is acknowledged as the case this cue misses.
- Topics carry their own prior for sarcasm, so the theme of a text shifts how likely an ironic reading is before any individual cue is examined.
- One way to classify is to run the generative model once under each candidate label and keep the label that explains the text best. This turns a generative model into a classifier without adding new parameters.
- Both baselines score nearly perfect recall with tiny precision, which shows they almost always predict "sarcastic"; reporting P and R separately makes this degenerate behaviour visible.

## For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

- **I1** (frames): prag could consult a precompiled "evaluative vs. descriptive" prior per lemma when deciding which nodes of a frame carry speaker attitude rather than event content.
- **I2** (frames): instead of counting words, the irony module can check polarity clash along dependency arcs, such as a positive predicate whose complement clause describes a bad state, and add a separate exaggeration rule for hyperbole.
- **I3** (world): the world model could tag a situation's domain (work overtime, bad weather) and pass a raised irony prior to prag, so the decision stays explainable as "domain prior + local clash".
- **I4** (answer): when an answer depends on a literal vs. ironic reading, Mova could build both interpretations of the text's world and choose the one more consistent with the rest of the story.
- **I5** (learn): evaluations of the irony detector should include an always-sarcastic negative control and report precision and recall separately, so a trivial labeller cannot pass as a good one.

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

📚 Shelves (4): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 9,086 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 8,889 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 9,296 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 9,206 works

🧭 Topics (1): [😊 Sentiment & emotion](https://md.research.elamur.ai/free/en/topics/sentiment-emotion/) · 608 works

## Cite

Cite as: Aditya Joshi, Prayas Jain, Pushpak Bhattacharyya and Mark Carman. 2016. ‘Who would have thought of that!': A Hierarchical Topic Model for Extraction of Sarcasm-prevalent Topics and Sarcasm Detection. Proceedings of the Workshop on Extra-Propositional Aspects of Meaning in Computational Linguistics (ExProM).

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{joshi2016who,
  title = {{‘Who would have thought of that!': A Hierarchical Topic Model for Extraction of Sarcasm-prevalent Topics and Sarcasm Detection}},
  author = {Aditya Joshi, Prayas Jain, Pushpak Bhattacharyya and Mark Carman},
  booktitle = {Proceedings of the Workshop on Extra-Propositional Aspects of Meaning in Computational Linguistics (ExProM)},
  year = {2016},
  eprint = {1611.04326},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1611.04326}
}
```

- [← previous Supplemental update](https://md.research.elamur.ai/free/en/papers/martin-2016-supplemental-update/)
- [next → Evaluating Inter-Annotator Agreement on Historical Spelling Normalization](https://md.research.elamur.ai/free/en/papers/bollmann-2016-evaluating-inter-annotator-agreement-historical/)
- [↑ back to papers of 2016](https://md.research.elamur.ai/free/en/papers/2016/page/4/)

## References

23 references, 12 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Stefano Baccianella, Andrea Esuli, and Fabrizio Sebastiani. 2010. [Sentiwordnet 3.0: An enhanced lexical resource for sentiment analysis and opinion mining](https://md.research.elamur.ai/free/en/papers/baccianella-2010-sentiwordnet-3-0-enhanced-lexical/). In LREC, volume 10, pages 2200–2204.
2. David Bamman and Noah A Smith. 2015. Contextualized sarcasm detection on twitter. In Ninth International AAAI Conference on Web and Social Media. [find](https://scholar.google.com/scholar?q=David+Bamman+and+Noah+A+Smith%2E+2015%2E+Contextualized+sarcasm+detection+on+twitter%2E+In+Ninth+International+AAAI+Conference+on+Web+and+Social+Media%2E)
3. David M Blei, Andrew Y Ng, and Michael I Jordan. 2003. [Latent dirichlet allocation](https://md.research.elamur.ai/free/en/papers/blei-2003-latent-dirichlet-allocation/). Journal of machine Learning research, 3(Jan):993–1022. [↗](https://doi.org/10.5555/944919.944937)
4. Konstantin Buschmeier, Philipp Cimiano, and Roman Klinger. 2014. [An impact analysis of features in a classification approach to irony detection in product reviews](https://md.research.elamur.ai/free/en/papers/buschmeier-2014-impact-analysis-features-classification-approach/). In Proceedings of the 5th Workshop on Computational Approaches to Subjectivity, Sentiment and Social Media Analysis, pages 42–49.
5. Yohan Jo and Alice H Oh. 2011. Aspect and sentiment unification model for online review analysis. In Proceedings of the fourth ACM international conference on Web search and data mining, pages 815–824. ACM. [find](https://scholar.google.com/scholar?q=Yohan+Jo+and+Alice+H+Oh%2E+2011%2E+Aspect+and+sentiment+unification+model+for+online+review+analysis%2E+In+Proceedings+of+the+fourth+ACM+international+conference+on+Web+search+and+data+mining%2C+pages+815%E2%80%93824)
6. Aditya Joshi, Vinita Sharma, and Pushpak Bhattacharyya. 2015. [Harnessing context incongruity for sarcasm detection](https://md.research.elamur.ai/free/en/papers/joshi-2015-harnessing-context-incongruity-sarcasm-detection/). In Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing, volume 2, pages 757–762.
7. Aditya Joshi, Pushpak Bhattacharyya, and Mark Carman. 2016a. Political issue extraction model: A novel hierarchical topic model that uses tweets by political and non-political authors. In 7th Workshop on Computational Approaches to Subjectivity, Sentiment & Social Media Analysis, pages 82–90. [find](https://scholar.google.com/scholar?q=Aditya+Joshi%2C+Pushpak+Bhattacharyya%2C+and+Mark+Carman%2E+2016a%2E+Political+issue+extraction+model%3A+A+novel+hierarchical+topic+model+that+uses+tweets+by+political+and+non%2Dpolitical+authors%2E+In+7th+Workshop)
8. Aditya Joshi, Pushpak Bhattacharyya, and Mark James Carman. 2016b. [Automatic sarcasm detection: A survey](https://md.research.elamur.ai/free/en/papers/joshi-2016-automatic-sarcasm-detection-survey/). arXiv preprint arXiv:1602.03426. [↗](https://arxiv.org/abs/1602.03426)
9. Aditya Joshi, Vaibhav Tripathi, Pushpak Bhattacharyya, and Mark Carman. 2016c. [Harnessing sequence labeling for sarcasm detection in dialogue from tv series ‘friends’](https://md.research.elamur.ai/free/en/papers/joshi-2016-harnessing-sequence-labeling-sarcasm-detection/). CoNLL 2016, page 146.
10. Anupam Khattri, Aditya Joshi, Pushpak Bhattacharyya, and Mark James Carman. 2015. [Your sentiment precedes you: Using an author’s historical tweets to predict sarcasm](https://md.research.elamur.ai/free/en/papers/khattri-2015-your-sentiment-precedes-you-author/). In 6TH WORKSHOP ON COMPUTATIONAL APPROACHES TO SUBJECTIVITY, SENTIMENT AND SOCIAL MEDIA ANALYSIS WASSA 2015, page 25. [↗](https://doi.org/10.60692/6y9tk-x1613)
11. Suin Kim, Jianwen Zhang, Zheng Chen, Alice H Oh, and Shixia Liu. 2013. A hierarchical aspect-sentiment model for online reviews. In AAAI. [find](https://scholar.google.com/scholar?q=Suin+Kim%2C+Jianwen+Zhang%2C+Zheng+Chen%2C+Alice+H+Oh%2C+and+Shixia+Liu%2E+2013%2E+A+hierarchical+aspect%2Dsentiment+model+for+online+reviews%2E+In+AAAI%2E)
12. CC Liebrecht, FA Kunneman, and APJ van den Bosch. 2013. [The perfect solution for detecting sarcasm in tweets# not](https://md.research.elamur.ai/free/en/papers/liebrecht-2013-perfect-solution-detecting-sarcasm-tweets/).
13. Julian McAuley and Jure Leskovec. 2013a. Hidden factors and hidden topics: understanding rating dimensions with review text. In Proceedings of the 7th ACM conference on Recommender systems, pages 165–172. ACM. [find](https://scholar.google.com/scholar?q=Julian+McAuley+and+Jure+Leskovec%2E+2013a%2E+Hidden+factors+and+hidden+topics%3A+understanding+rating+dimensions+with+review+text%2E+In+Proceedings+of+the+7th+ACM+conference+on+Recommender+systems%2C+pages+165%E2%80%93)
14. Julian John McAuley and Jure Leskovec. 2013b. From amateurs to connoisseurs: modeling the evolution of user expertise through online reviews. In Proceedings of the 22nd international conference on World Wide Web, pages 897–908. International World Wide Web Conferences Steering Committee. [↗](https://doi.org/10.48550/arxiv.1303.4402)
15. Arjun Mukherjee and Bing Liu. 2012a. Aspect extraction through semi-supervised modeling. In Proceedings of the 50th Annual Meeting of the Association for Computational Linguistics: Long Papers-Volume 1, pages 339–348. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Arjun+Mukherjee+and+Bing+Liu%2E+2012a%2E+Aspect+extraction+through+semi%2Dsupervised+modeling%2E+In+Proceedings+of+the+50th+Annual+Meeting+of+the+Association+for+Computational+Linguistics%3A+Long+Papers%2DVolume+)
16. Arjun Mukherjee and Bing Liu. 2012b. Aspect extraction through semi-supervised modeling. In Proceedings of the 50th Annual Meeting of the Association for Computational Linguistics: Long Papers-Volume 1, pages 339–348. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Arjun+Mukherjee+and+Bing+Liu%2E+2012b%2E+Aspect+extraction+through+semi%2Dsupervised+modeling%2E+In+Proceedings+of+the+50th+Annual+Meeting+of+the+Association+for+Computational+Linguistics%3A+Long+Papers%2DVolume+)
17. Arjun Mukherjee and Bing Liu. 2012c. Modeling review comments. In Proceedings of the 50th Annual Meeting of the Association for Computational Linguistics: Long Papers-Volume 1, pages 320–329. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Arjun+Mukherjee+and+Bing+Liu%2E+2012c%2E+Modeling+review+comments%2E+In+Proceedings+of+the+50th+Annual+Meeting+of+the+Association+for+Computational+Linguistics%3A+Long+Papers%2DVolume+1%2C+pages+320%E2%80%93329%2E+Associat)
18. Ashwin Rajadesingan, Reza Zafarani, and Huan Liu. 2015. Sarcasm detection on twitter: A behavioral modeling approach. In Proceedings of the Eighth ACM International Conference on Web Search and Data Mining, WSDM ’15, pages 97–106, New York, NY, USA. ACM. [find](https://scholar.google.com/scholar?q=Ashwin+Rajadesingan%2C+Reza+Zafarani%2C+and+Huan+Liu%2E+2015%2E+Sarcasm+detection+on+twitter%3A+A+behavioral+modeling+approach%2E+In+Proceedings+of+the+Eighth+ACM+International+Conference+on+Web+Search+and+Data+M)
19. Daniel Ramage, David Hall, Ramesh Nallapati, and Christopher D. Manning. 2009. [Labeled lda: A supervised topic model for credit attribution in multi-labeled corpora](https://md.research.elamur.ai/free/en/papers/ramage-2009-labeled-lda-supervised-topic-model/). In Proceedings of the 2009 Conference on Empirical Methods in Natural Language Processing: Volume 1 - Volume 1, EMNLP ’09, pages 248–256, Stroudsburg, PA, USA. Association for Computational Linguistics.
20. Ellen Riloff, Ashequl Qadir, Prafulla Surve, Lalindra De Silva, Nathan Gilbert, and Ruihong Huang. 2013. [Sarcasm as contrast between a positive sentiment and negative situation](https://md.research.elamur.ai/free/en/papers/riloff-2013-sarcasm-as-contrast-between-positive/). In EMNLP, volume 13, pages 704–714.
21. Byron C Silvio Amir, Wallace, Hao Lyu, and Paula Carvalho Mário J Silva. 2016. [Modelling context with user embeddings for sarcasm detection in social media](https://md.research.elamur.ai/free/en/papers/amir-2016-modelling-context-user-embeddings-sarcasm/). CoNLL 2016, page 167.
22. Byron C. Wallace, Do Kook Choe, Laura Kertz, and Eugene Charniak. 2014. [Humans require context to infer ironic intent (so computers probably do, too)](https://md.research.elamur.ai/free/en/papers/wallace-2014-humans-require-context-infer-ironic/). In Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pages 512–516, Baltimore, Maryland, June. Association for Computational Linguistics.
23. Zelin Wang, Zhijian Wu, Ruimin Wang, and Yafeng Ren. 2015. Twitter sarcasm detection exploiting a context-based model. In Web Information Systems Engineering–WISE 2015, pages 77–91. Springer. [find](https://scholar.google.com/scholar?q=Zelin+Wang%2C+Zhijian+Wu%2C+Ruimin+Wang%2C+and+Yafeng+Ren%2E+2015%2E+Twitter+sarcasm+detection+exploiting+a+context%2Dbased+model%2E+In+Web+Information+Systems+Engineering%E2%80%93WISE+2015%2C+pages+77%E2%80%9391%2E+Springer%2E)

*Updated 24 October 2026*