---
title: "A Weak Supervised Transfer Learning Approach for Sentiment"
description: "This design follows Smith et al. (2022). 93% accuracy. 36 references. Among them: 13."
authors: ["Fatemah Husain", "Hana Al-Ostad", "Halima Omar"]
year: 2022
venue: "Proceedings of the Seventh Arabic Natural Language Processing Workshop (WANLP)"
source: "https://aclanthology.org/2022.wanlp-1.15/"
doi: "10.18653/v1/2022.wanlp-1.15"
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/husain-2022-weak-supervised-transfer-learning-approach/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/husain-2022-weak-supervised-transfer-learning-approach/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2022](https://md.research.elamur.ai/free/en/papers/2022/page/6/) / A Weak Supervised Transfer Learning Approach for Sentiment Analysis to the Kuwaiti Dialect

# A Weak Supervised Transfer Learning Approach for Sentiment Analysis to the Kuwaiti Dialect

**Authors:** [Fatemah Husain](https://md.research.elamur.ai/free/en/authors/fatemah-husain/), Hana Al-Ostad and Halima Omar · **Year:** 2022 · **Venue:** [Proceedings of the Seventh Arabic Natural Language Processing Workshop (WANLP)](https://md.research.elamur.ai/free/en/venues/wanlp/) ([WANLP 2022](https://md.research.elamur.ai/free/en/venues/wanlp/2022/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2022.wanlp-1.15/) · [PDF](https://aclanthology.org/2022.wanlp-1.15.pdf) · [DOI 10.18653/v1/2022.wanlp-1.15](https://doi.org/10.18653/v1/2022.wanlp-1.15) · [OpenAlex](https://openalex.org/W4385573024)

**Citations:** 13 citations (OpenAlex, 2026-10-14)

## Summary
Arabic sentiment analysis has few datasets, and many dialects, Kuwaiti among them, have almost none. A major obstacle is the cost and difficulty of recruiting qualified annotators. The authors address this by building a large labeled Kuwaiti tweet corpus with very little manual annotation. Their labeling pipeline uses Snorkel. Each labeling function sends a prompt to a pretrained zero-shot NLI model, and Snorkel's label model combines and denoises the resulting votes into probabilistic labels. This design follows Smith et al. (2022). The final corpus has more than 16.6k tweets drawn from various topics and time periods. A human-labeled subset measures how good the automatic labels are, and agreement is high. The authors then train classical classifiers and several BERT variants on the data. ARBERT gives the most credible results. The work matters for two reasons. It is the first public Kuwaiti sentiment resource, and its labeling recipe could carry over to related Gulf dialects such as Bahraini or Qatari.

## Key points
- The dataset contains 7,905 negative, 7,902 positive and 860 neutral tweets. Personal and sensitive content was removed, and fluent Kuwaiti annotators labeled the evaluation set by hand.
- Snorkel's labels agree with the human gold labels at 93% pairwise and reach Cohen's kappa 0.87. On a gold test set of 1,534 tweets, the labeler scores 93% accuracy and 84% macro-F1.
- On the test set, ARBERT reaches 0.75 macro-F1 and 0.89 accuracy. It beats AraBERT, MARBERT and MiniLM, even though AraBERT and MARBERT were pretrained on tweet-like data.
- The SVM and bagging baselines score almost perfectly (SVM 0.99 F1 and 1.00 accuracy). The authors read these numbers as likely overfitting rather than real success.
- The class distribution is skewed, so the authors report macro-averaged metrics with stratified 5-fold cross-validation. Error analysis shows many ambiguous tweets and frequent English borrowings written in Arabic script.

## Interesting ideas
- Each labeling function wraps a zero-shot NLI prompt, and a label model reconciles the noisy votes, so no hand-labeled training data is needed.
- A small human-labeled set measures the automatic labeler itself, using both agreement and kappa, before anyone trusts the downstream models.
- The authors treat near-perfect classifier scores as a warning of overfitting rather than as a success.
- Error analysis separates explicitly polar tweets from ambiguous ones whose topic or stance is unclear.
- Dialect text keeps absorbing English loanwords respelled in the local script.

## For Mova

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

- **I1** (learn): Prompted NLI checkers could serve as extra voting witnesses next to the parsers when producing silver data, with a label model estimating how reliable each witness is.
- **I2** (learn): Every silver-data generator could be scored against a small gold slice first, so that weak labels feeding training come with known reliability.
- **I3** (learn): Evaluation could automatically flag implausibly high scores and pair them with negative controls before a model is promoted.
- **I4** (frames): Pragmatics evaluation could tag items as explicit or implied attitude, so irony and implicature failures are counted separately from plain misreadings.
- **I5** (tokens): The lexicon and lemmatizer could keep a recognizable class for respelled borrowings and nonstandard forms instead of treating them as unknown noise.

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

## Institutions

- [Kuwait University](https://md.research.elamur.ai/free/en/universities/kuwait-university/) — University, Kuwait City, Kuwait · 11 works · 147 citations

Also on the byline: [Gulf University for Science & Technology](https://ror.org/04d9rzd67).

📚 Shelves (4): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 16,667 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 16,194 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 16,039 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 15,891 works

🧭 Topics (2): [😊 Sentiment & emotion](https://md.research.elamur.ai/free/en/topics/sentiment-emotion/) · 972 works · [Semitic languages](https://md.research.elamur.ai/free/en/topics/semitic-languages/) · 2,908 works

## Cite

Cite as: Fatemah Husain, Hana Al-Ostad and Halima Omar. 2022. A Weak Supervised Transfer Learning Approach for Sentiment Analysis to the Kuwaiti Dialect. Proceedings of the Seventh Arabic Natural Language Processing Workshop (WANLP).

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

```bibtex
@inproceedings{husain2022weak,
  title = {{A Weak Supervised Transfer Learning Approach for Sentiment Analysis to the Kuwaiti Dialect}},
  author = {Fatemah Husain, Hana Al-Ostad and Halima Omar},
  booktitle = {Proceedings of the Seventh Arabic Natural Language Processing Workshop (WANLP)},
  year = {2022},
  doi = {10.18653/v1/2022.wanlp-1.15},
  url = {https://doi.org/10.18653/v1/2022.wanlp-1.15}
}
```

- [← previous Uncertainty and Inclusivity in Gender Bias Annotation: An Annotation Taxonomy and Annotated Datasets of British English Text](https://md.research.elamur.ai/free/en/papers/havens-2022-uncertainty-inclusivity-gender-bias-annotation/)
- [next → ArCovidVac: Analyzing Arabic Tweets About COVID-19 Vaccination](https://md.research.elamur.ai/free/en/papers/mubarak-2022-arcovidvac-analyzing-arabic-tweets-about/)
- [↑ back to papers of 2022](https://md.research.elamur.ai/free/en/papers/2022/page/6/)

## References

36 references, 7 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Arwa A. Al Shamsi and Sherief Abdallah. 2022. Sentiment analysis of emirati dialect. Big Data and Cognitive Computing, 6(2). [↗](https://doi.org/10.3390/bdcc6020057)
2. Muhammad Abdul-Mageed, AbdelRahim Elmadany, and El Moatez Billah Nagoudi. 2021. [ARBERT & MARBERT: Deep bidirectional transformers for Arabic](https://md.research.elamur.ai/free/en/papers/abdul-mageed-2021-arbert-marbert-deep-bidirectional-transformers/). In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers), pages 7088–7105, Online. Association for Computational Linguistics. [↗](https://doi.org/10.18653/v1/2021.acl-long.551)
3. Snorkel AI. Advancing snorkel from research to production. Last accessed 02 September 2022. [find](https://scholar.google.com/scholar?q=Snorkel+AI%2E+Advancing+snorkel+from+research+to+production%2E+Last+accessed+02+September+2022%2E)
4. Shamlan Al-Qenaie et al. 2011. Kuwaiti Arabic: A socio-phonological perspective. Ph.D. thesis, Durham University. [find](https://scholar.google.com/scholar?q=Shamlan+Al%2DQenaie+et+al%2E+2011%2E+Kuwaiti+Arabic%3A+A+socio%2Dphonological+perspective%2E+Ph%2ED%2E+thesis%2C+Durham+University%2E)
5. Rahma M. Alahmary, Hmood Z. Al-Dossari, and Ahmed Z. Emam. 2019. Sentiment analysis of saudi dialect using deep learning techniques. In 2019 International Conference on Electronics, Information, and Communication (ICEIC), pages 1–6. [find](https://scholar.google.com/scholar?q=Rahma+M%2E+Alahmary%2C+Hmood+Z%2E+Al%2DDossari%2C+and+Ahmed+Z%2E+Emam%2E+2019%2E+Sentiment+analysis+of+saudi+dialect+using+deep+learning+techniques%2E+In+2019+International+Conference+on+Electronics%2C+Information%2C+and+C)
6. Banan A. Alharbi, Mohammad A. Mezher, and Abdullah M. Barakeh. 2022. Tourist reviews sentiment classification using deep learning techniques: A case study in saudi arabia. International Journal of Advanced Computer Science and Applications, 13(6). [↗](https://doi.org/10.14569/ijacsa.2022.0130685)
7. Meshrif Alruily and Osama R Shahin. 2020. Sentiment analysis of twitter data for saudi universities. International Journal of Machine Learning and Computing, 10(1). [↗](https://doi.org/10.18178/ijmlc.2020.10.1.892)
8. Wissam Antoun, Fady Baly, and Hazem Hajj. [Arabert: Transformer-based model for arabic language understanding](https://md.research.elamur.ai/free/en/papers/antoun-2020-arabert-transformer-based-model-arabic/). In LREC 2020 Workshop Language Resources and Evaluation Conference 11–16 May 2020, page 9. [↗](https://doi.org/10.48550/arxiv.2003.00104)
9. Aditya Bohra, Deepanshu Vijay, Vinay Singh, Syed Sarfaraz Akhtar, and Manish Shrivastava. 2018. [A dataset of hindi-english code-mixed social media text for hate speech detection](https://md.research.elamur.ai/free/en/papers/bohra-2018-dataset-hindi-english-code-mixed/). In Proceedings of the second workshop on computational modeling of people’s opinions, personality, and emotions in social media, pages 36–41.
10. Alexis Conneau, Kartikay Khandelwal, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzmán, Edouard Grave, Myle Ott, Luke Zettlemoyer, and Veselin Stoyanov. 2020. Unsupervised cross-lingual representation learning at scale. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 8440– 8451, Online. Association for Computational Linguistics. Joe Davison. xlm-roberta-large-xnli. [find](https://scholar.google.com/scholar?q=Alexis+Conneau%2C+Kartikay+Khandelwal%2C+Naman+Goyal%2C+Vishrav+Chaudhary%2C+Guillaume+Wenzek%2C+Francisco+Guzm%C3%A1n%2C+Edouard+Grave%2C+Myle+Ott%2C+Luke+Zettlemoyer%2C+and+Veselin+Stoyanov%2E+2020%2E+Unsupervised+cross%2Dlingu)
11. Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2018. [Bert: Pre-training of deep bidirectional transformers for language understanding](https://md.research.elamur.ai/free/en/papers/devlin-2019-bert-pre-training-deep-bidirectional/). arXiv preprint arXiv:1810.04805. [↗](https://arxiv.org/abs/1810.04805)
12. Jesper E. van Engelen and Holger H. Hoos. 2020. A survey on semi-supervised learning. Machine Learning, 109(2):373–440. [↗](https://doi.org/10.1007/s10994-019-05855-6)
13. Jason Fries, Sen Wu, Alex Ratner, and Christopher Ré. 2017. Swellshark: A generative model for biomedical named entity recognition without labeled data. arXiv preprint arXiv:1704.06360. [↗](https://arxiv.org/abs/1704.06360)
14. Daniel Fu, Mayee Chen, Frederic Sala, Sarah Hooper, Kayvon Fatahalian, and Christopher Ré. 2020. Fast and three-rious: Speeding up weak supervision with triplet methods. In International Conference on Machine Learning, pages 3280–3291. PMLR. [↗](https://doi.org/10.48550/arxiv.2002.11955)
15. Nizar Y. Habash. 2010. 1 edition, volume 3 of Synthesis Lectures on Human Language Technologies. Morgan Claypool Publishers. [link]. [find](https://scholar.google.com/scholar?q=Nizar+Y%2E+Habash%2E+2010%2E+1+edition%2C+volume+3+of+Synthesis+Lectures+on+Human+Language+Technologies%2E+Morgan+Claypool+Publishers%2E+%5Blink%5D%2E)
16. Noor A Hayat and Yousuf B AlBader. 2022. The mcchicken phenomenon: How has english become a prevalent language among kuwaiti youths? World Journal of English Language, 12(6). [↗](https://doi.org/10.5430/wjel.v12n6p59)
17. Pengcheng He, Jianfeng Gao, and Weizhu Chen. 2021. [DeBERTaV3: Improving DeBERTa using ELECTRA-Style Pre-Training with Gradient-Disentangled Embedding Sharing](https://md.research.elamur.ai/free/en/papers/he-2021-debertav3-improving-deberta-using-electra/). arXiv. [↗](https://doi.org/10.48550/arxiv.2111.09543)
18. Fatemah Husain. 2020. [OSACT4 shared task on offensive language detection: Intensive preprocessing-based approach](https://md.research.elamur.ai/free/en/papers/husain-2020-osact4-shared-task-offensive-language/). In Proceedings of the 4th Workshop on Open-Source Arabic Corpora and Processing Tools, with a Shared Task on Offensive Language Detection, pages 53–60, Marseille, France. European Language Resource Association. [↗](https://doi.org/10.48550/arxiv.2005.07297)
19. Fatemah Husain and Ozlem Uzuner. 2021. A survey of offensive language detection for the arabic language. ACM Trans. Asian Low-Resour. Lang. Inf. Process., 20(1). [↗](https://doi.org/10.1145/3421504)
20. Fatemah Husain and Ozlem Uzuner. 2022. Investigating the effect of preprocessing arabic text on offensive language and hate speech detection. ACM Trans. Asian Low-Resour. Lang. Inf. Process., 21(4). [↗](https://doi.org/10.1145/3501398)
21. Moritz Laurer, Wouter van Atteveldt, Andreu Casas, and Kasper Welbers. 2022. Less annotating, more classifying – addressing the data scarcity issue of supervised machine learning with deep transfer learning and bert-nli. [↗](https://doi.org/10.1017/pan.2023.20)
22. Chikashi Nobata, Joel Tetreault, Achint Thomas, Yashar Mehdad, and Yi Chang. 2016. Abusive language detection in online user content. In Proceedings of the 25th international conference on world wide web, pages 145–153. [find](https://scholar.google.com/scholar?q=Chikashi+Nobata%2C+Joel+Tetreault%2C+Achint+Thomas%2C+Yashar+Mehdad%2C+and+Yi+Chang%2E+2016%2E+Abusive+language+detection+in+online+user+content%2E+In+Proceedings+of+the+25th+international+conference+on+world+wide+)
23. Thuraya M Omran, Baraa T Sharef, Crina Grosan, and Yongmin Li. 2022. Transfer Learning and Sentiment Analysis of Bahraini Dialects Sequential Text Data using Multilingual Deep Learning Approach. Chaos, Solitons and Fractals. [↗](https://doi.org/10.1016/j.datak.2022.102106)
24. Sinno Jialin Pan and Qiang Yang. 2010. A survey on transfer learning. IEEE Transactions on Knowledge and Data Engineering, 22(10):1345–1359. [↗](https://doi.org/10.1109/tkde.2009.191)
25. A. Ratner, P. Varma, B. Hancock, and C. Ré. Weak supervision: A new programming paradigm for machine learning. Last accessed 02 September 2022. [find](https://scholar.google.com/scholar?q=A%2E+Ratner%2C+P%2E+Varma%2C+B%2E+Hancock%2C+and+C%2E+R%C3%A9%2E+Weak+supervision%3A+A+new+programming+paradigm+for+machine+learning%2E+Last+accessed+02+September+2022%2E)
26. Alexander Ratner, Stephen H Bach, Henry Ehrenberg, Jason Fries, Sen Wu, and Christopher Ré. 2017. Snorkel: rapid training data creation with weak supervision. Proceedings of the VLDB Endowment, 11(3):269–282. [find](https://scholar.google.com/scholar?q=Alexander+Ratner%2C+Stephen+H+Bach%2C+Henry+Ehrenberg%2C+Jason+Fries%2C+Sen+Wu%2C+and+Christopher+R%C3%A9%2E+2017%2E+Snorkel%3A+rapid+training+data+creation+with+weak+supervision%2E+Proceedings+of+the+VLDB+Endowment%2C+11%283%29%3A)
27. Alexander J Ratner, Christopher M De Sa, Sen Wu, Daniel Selsam, and Christopher Ré. 2016. Data programming: Creating large training sets, quickly. Advances in neural information processing systems, 29. [find](https://scholar.google.com/scholar?q=Alexander+J+Ratner%2C+Christopher+M+De+Sa%2C+Sen+Wu%2C+Daniel+Selsam%2C+and+Christopher+R%C3%A9%2E+2016%2E+Data+programming%3A+Creating+large+training+sets%2C+quickly%2E+Advances+in+neural+information+processing+systems%2C+29)
28. Sandra Rizkallah, Amir Atiya, Hossam ElDin Mahgoub, and Momen Heragy. 2018. Dialect versus msa sentiment analysis. In The International Conference on Advanced Machine Learning Technologies and Applications (AMLTA2018), pages 605–613, Cham. Springer International Publishing. [find](https://scholar.google.com/scholar?q=Sandra+Rizkallah%2C+Amir+Atiya%2C+Hossam+ElDin+Mahgoub%2C+and+Momen+Heragy%2E+2018%2E+Dialect+versus+msa+sentiment+analysis%2E+In+The+International+Conference+on+Advanced+Machine+Learning+Technologies+and+Applica)
29. Joshua Robinson, Stefanie Jegelka, and Suvrit Sra. 2020. Strength from weakness: Fast learning using weak supervision. In Proceedings of the 37th International Conference on Machine Learning, volume 119 of Proceedings of Machine Learning Research, pages 8127–8136. PMLR. [↗](https://doi.org/10.48550/arxiv.2002.08483)
30. Janan Ben Salamah and Aymen Elkhlifi. 2014. Microblogging opinion mining approach for kuwaiti dialect. In The International Conference on Computing Technology and Information Management (ICCTIM), page 388. Citeseer. [find](https://scholar.google.com/scholar?q=Janan+Ben+Salamah+and+Aymen+Elkhlifi%2E+2014%2E+Microblogging+opinion+mining+approach+for+kuwaiti+dialect%2E+In+The+International+Conference+on+Computing+Technology+and+Information+Management+%28ICCTIM%29%2C+page)
31. Burr Settles. 2009. [Active learning literature survey](https://md.research.elamur.ai/free/en/papers/settles-2009-active-learning-literature-survey/).
32. Ryan Smith, Jason A Fries, Braden Hancock, and Stephen H Bach. 2022. Language Models in the Loop: Incorporating Prompting into Weak Supervision. arXiv. [find](https://scholar.google.com/scholar?q=Ryan+Smith%2C+Jason+A+Fries%2C+Braden+Hancock%2C+and+Stephen+H+Bach%2E+2022%2E+Language+Models+in+the+Loop%3A+Incorporating+Prompting+into+Weak+Supervision%2E+arXiv%2E)
33. W.H. Tok, A. Bahree, and S. Filipi. 2021. Practical Weak Supervision: Doing More with Less Data. O’Reilly Media, Incorporated. [find](https://scholar.google.com/scholar?q=W%2EH%2E+Tok%2C+A%2E+Bahree%2C+and+S%2E+Filipi%2E+2021%2E+Practical+Weak+Supervision%3A+Doing+More+with+Less+Data%2E+O%E2%80%99Reilly+Media%2C+Incorporated%2E)
34. Lewis Tunstall, Leandro von Werra, and Thomas Wolf. 2022. Natural language processing with transformers. " O’Reilly Media, Inc.". [find](https://scholar.google.com/scholar?q=Lewis+Tunstall%2C+Leandro+von+Werra%2C+and+Thomas+Wolf%2E+2022%2E+Natural+language+processing+with+transformers%2E+%22+O%E2%80%99Reilly+Media%2C+Inc%2E%22%2E)
35. Wenhui Wang, Furu Wei, Li Dong, Hangbo Bao, Nan Yang, and Ming Zhou. 2020. Minilm: Deep self-attention distillation for task-agnostic compression of pre-trained transformers. [find](https://scholar.google.com/scholar?q=Wenhui+Wang%2C+Furu+Wei%2C+Li+Dong%2C+Hangbo+Bao%2C+Nan+Yang%2C+and+Ming+Zhou%2E+2020%2E+Minilm%3A+Deep+self%2Dattention+distillation+for+task%2Dagnostic+compression+of+pre%2Dtrained+transformers%2E)
36. S. Yildirim and M. Asgari-Chenaghlu. 2021. Mastering Transformers: Build state-of-the-art models from scratch with advanced natural language processing techniques. Packt Publishing. [find](https://scholar.google.com/scholar?q=S%2E+Yildirim+and+M%2E+Asgari%2DChenaghlu%2E+2021%2E+Mastering+Transformers%3A+Build+state%2Dof%2Dthe%2Dart+models+from+scratch+with+advanced+natural+language+processing+techniques%2E+Packt+Publishing%2E)

*Updated 1 November 2026*