---
title: "SalamNET at SemEval-2020 Task 12: Deep Learning Approach"
description: "The paper describes SalamNET, a system for detecting offensive Arabic tweets, built for… 21 references, cited by 3 papers here."
authors: ["Fatemah Husain", "Jooyeon Lee", "Sam Henry", "Ozlem Uzuner"]
year: 2020
venue: "Proceedings of the Fourteenth Workshop on Semantic Evaluation"
source: "https://aclanthology.org/2020.semeval-1.283/"
doi: "10.18653/v1/2020.semeval-1.283"
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-2020-salamnet-semeval-2020-task-12/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/husain-2020-salamnet-semeval-2020-task-12/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2020](https://md.research.elamur.ai/free/en/papers/2020/page/7/) / SalamNET at SemEval-2020 Task 12: Deep Learning Approach for Arabic Offensive Language Detection

# SalamNET at SemEval-2020 Task 12: Deep Learning Approach for Arabic Offensive Language Detection

**Authors:** [Fatemah Husain](https://md.research.elamur.ai/free/en/authors/fatemah-husain/), Jooyeon Lee, [Sam Henry](https://md.research.elamur.ai/free/en/authors/sam-henry/) and [Ozlem Uzuner](https://md.research.elamur.ai/free/en/authors/ozlem-uzuner/) · **Year:** 2020 · **Venue:** [Proceedings of the Fourteenth Workshop on Semantic Evaluation](https://md.research.elamur.ai/free/en/venues/semeval/) ([SemEval 2020](https://md.research.elamur.ai/free/en/venues/semeval/2020/)) · **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/2020.semeval-1.283/) · [PDF](https://aclanthology.org/2020.semeval-1.283.pdf) · [DOI 10.18653/v1/2020.semeval-1.283](https://doi.org/10.18653/v1/2020.semeval-1.283) · [OpenAlex](https://openalex.org/W3045993639)

**Citations:** 14 citations (OpenAlex, 2026-10-14) · 3 in our library, of the 149,439 papers whose bibliographies we read

## Summary
The paper describes SalamNET, a system for detecting offensive Arabic tweets, built for SemEval-2020 Task 12 (OffensEval). The authors used the Arabic OffensEval 2020 dataset: 10,000 tweets, of which only about 19% are offensive, split 7,000/1,000/2,000 for training, development and test. Tweets went through heavy Arabic-specific preprocessing. This included converting emoji to text labels, normalizing letter variants, mapping dialectal nouns to MSA, replacing hyponyms such as animal names with a hypernym, splitting hashtags, and general cleaning. The authors then compared a Logistic Regression baseline with RNN, GRU, Bi-GRU, LSTM and Bi-LSTM models, each fed either TF-IDF features or AraVec embeddings. Their official runs ranked only mid-table. In later experiments, a Bi-GRU with TF-IDF features reached a macro-F1 of 0.83. The most useful part of the paper is its error analysis. It shows where all the models fail together, which feature types suit which tweets, and why combining models and adding NER, dialect and religion-aware features could help.

## Key points
- The data is the Arabic OffensEval 2020 corpus: 10,000 tweets with 1,900 labeled offensive, using the official 7k/1k/2k split.
- The official submissions scored as follows, out of 53 teams, against a top score of 0.901:
  - LR with AraVec: macro-F1 0.81, 34th place.
  - Bi-GRU with TF-IDF: macro-F1 0.80, 38th place.
- The post-competition best was a Bi-GRU with TF-IDF at 0.83 macro-F1 (weighted-F1 0.84), evaluated with 10-fold cross-validation and grid-searched hyperparameters.
- Feature type interacted with model type:
  - AraVec (Twitter, CBOW) beat TF-IDF for LR (0.81 vs 0.71).
  - TF-IDF beat AraVec for every neural model.
  - Upsampling the minority class hurt performance and was dropped from the pipeline.
- On the test set, all models together missed 390 offensive tweets but wrongly flagged only 11 clean ones. Many of these shared failures were Egyptian-dialect sports tweets that mention famous people.

## Interesting ideas
- Collect the items that every model gets wrong and inspect them as a group, because errors shared by all models point to gaps in the data or representation rather than to one model's quirks.
- Compare the two feature families directly by studying cases one gets right and the other gets wrong, then describe those cases by length, vocabulary breadth or dialect.
- Replacing hyponyms with a shared hypernym (e.g., any animal name → "animal") reduces sparsity, but it can erase distinctions that carry meaning. The authors therefore suggest splitting animal categories by whether they are used offensively.
- Normalize dialectal and orthographic variants to a standard form before modeling, while noticing that the standard form itself may end up underrepresented.
- The analysis found that mentioning a sensitive topic, such as a prayer or a religious figure, is different from attacking someone, and topic words alone blur the two.

## For Mova

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

- **I1** (learn): Run all witness parsers on the same text and route sentences where all of them fail into a separate hard-case pool for LLM-teacher review or treebank fixes.
- **I2** (tree): Build perceptron-vs-gparse disagreement sets and profile them by sentence length or construction to decide where each parser should be trusted.
- **I3** (world): When backing off rare nouns to WordNet hypernyms in the ontology, keep finer subclasses whenever they carry evaluative or connotative meaning.
- **I4** (tokens): Map informal spellings and regional variants to standard lemmas in crate en, but keep the original surface form so the mapping stays explainable.
- **I5** (frames): Decide offensiveness or irony from who does what to whom in the frame and from the speech act, not from the presence of a topic word.

*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

- [George Mason University](https://md.research.elamur.ai/free/en/universities/george-mason-university/) — University, Fairfax, United States · 24 works · 1,288 citations
- [Kuwait University](https://md.research.elamur.ai/free/en/universities/kuwait-university/) — University, Kuwait City, Kuwait · 11 works · 147 citations

📚 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, Jooyeon Lee, Sam Henry and Ozlem Uzuner. 2020. SalamNET at SemEval-2020 Task 12: Deep Learning Approach for Arabic Offensive Language Detection. Proceedings of the Fourteenth Workshop on Semantic Evaluation.

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

```bibtex
@inproceedings{husain2020salamnet,
  title = {{SalamNET at SemEval-2020 Task 12: Deep Learning Approach for Arabic Offensive Language Detection}},
  author = {Fatemah Husain, Jooyeon Lee, Sam Henry and Ozlem Uzuner},
  booktitle = {Proceedings of the Fourteenth Workshop on Semantic Evaluation},
  year = {2020},
  doi = {10.18653/v1/2020.semeval-1.283},
  url = {https://doi.org/10.18653/v1/2020.semeval-1.283}
}
```

- [← previous Improving Arabic Text Categorization Using Transformer Training Diversification](https://md.research.elamur.ai/free/en/papers/chowdhury-2020-improving-arabic-text-categorization-transformer/)
- [next → Stress Test Evaluation of Transformer-based Models in Natural Language Understanding Tasks](https://md.research.elamur.ai/free/en/papers/aspillaga-2020-stress-test-evaluation-transformer-based/)
- [↑ back to papers of 2020](https://md.research.elamur.ai/free/en/papers/2020/page/7/)

## References

21 references, 8 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Ehab A. Abozinadah and James H. Jones. 2017. A statistical learning approach to detect abusive twitter accounts. In Proceedings of the International Conference on Compute and Data Analysis, ICCDA ’17, page 6–13, New York, NY, USA. Association for Computing Machinery. [find](https://scholar.google.com/scholar?q=Ehab+A%2E+Abozinadah+and+James+H%2E+Jones%2E+2017%2E+A+statistical+learning+approach+to+detect+abusive+twitter+accounts%2E+In+Proceedings+of+the+International+Conference+on+Compute+and+Data+Analysis%2C+ICCDA+%E2%80%9917%2C)
2. Ibrahim Abu Farha and Walid Magdy. 2019. [Mazajak: An online Arabic sentiment analyser](https://md.research.elamur.ai/free/en/papers/farha-2019-mazajak-online-arabic-sentiment-analyser/). In Proceedings of the Fourth Arabic Natural Language Processing Workshop, pages 192–198, Florence, Italy, August. Association for Computational Linguistics.
3. Azalden Alakrot, Liam Murray, and Nikola Nikolov. 2018a. Dataset construction for the detection of anti-social behaviour in online communication in arabic. Procedia Computer Science, 142:174–181, 11. [find](https://scholar.google.com/scholar?q=Azalden+Alakrot%2C+Liam+Murray%2C+and+Nikola+Nikolov%2E+2018a%2E+Dataset+construction+for+the+detection+of+anti%2Dsocial+behaviour+in+online+communication+in+arabic%2E+Procedia+Computer+Science%2C+142%3A174%E2%80%93181%2C+11%2E)
4. Azalden Alakrot, Liam Murray, and Nikola S. Nikolov. 2018b. Dataset construction for the detection of anti-social behaviour in online communication in arabic. Procedia Computer Science, 142:174–181. [find](https://scholar.google.com/scholar?q=Azalden+Alakrot%2C+Liam+Murray%2C+and+Nikola+S%2E+Nikolov%2E+2018b%2E+Dataset+construction+for+the+detection+of+anti%2Dsocial+behaviour+in+online+communication+in+arabic%2E+Procedia+Computer+Science%2C+142%3A174%E2%80%93181%2E)
5. Nuha Albadi, Maram Kurdi, and Shivakant Mishra. 2018. Are they our brothers? analysis and detection of religious hate speech in the arabic twittersphere. In 2018 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM), pages 69–76, August. [find](https://scholar.google.com/scholar?q=Nuha+Albadi%2C+Maram+Kurdi%2C+and+Shivakant+Mishra%2E+2018%2E+Are+they+our+brothers%3F+analysis+and+detection+of+religious+hate+speech+in+the+arabic+twittersphere%2E+In+2018+IEEE%2FACM+International+Conference+on+A)
6. Nuha Albadi, Maram Kurdi, and Shivakant Mishra. 2019. Investigating the effect of combining GRU neural networks with handcrafted features for religious hatred detection on arabic twitter space. Social Network Analysis and Mining, 9(41):1–19, August. [↗](https://doi.org/10.1007/s13278-019-0587-5)
7. Ali Alshehri, El Moatez Billah Nagoudi, Hassan Alhuzali, and Muhammad Abdul-Mageed. 2018. Think before your click: Data and models for adult content in arabic twitter. In Second Workshop on Text Analytics for Cybersecurity and Online Safety (TA-COS 2018). European Language Resources Association (ELRA), May. [find](https://scholar.google.com/scholar?q=Ali+Alshehri%2C+El+Moatez+Billah+Nagoudi%2C+Hassan+Alhuzali%2C+and+Muhammad+Abdul%2DMageed%2E+2018%2E+Think+before+your+click%3A+Data+and+models+for+adult+content+in+arabic+twitter%2E+In+Second+Workshop+on+Text+Analy)
8. Piotr Bojanowski, Edouard Grave, Armand Joulin, and Tomas Mikolov. 2016. [Enriching word vectors with subword information](https://md.research.elamur.ai/free/en/papers/bojanowski-2017-enriching-word-vectors-subword-information/). cite arxiv:1607.04606Comment: Accepted to TACL. The two first authors contributed equally. [↗](https://arxiv.org/abs/1607.04606)
9. Arijit Ghosh Chowdhury, Aniket Didolkar, Ramit Sawhney, and Rajiv Ratn Shah. 2019. [ARHNet -leveraging community interaction for detection of religious hate speech in Arabic](https://md.research.elamur.ai/free/en/papers/chowdhury-2019-arhnet-leveraging-community-interaction-detection/). In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics: Student Research Workshop, pages 273–280, Florence, Italy, July. Association for Computational Linguistics.
10. Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. [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/). In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), pages 4171–4186, Minneapolis, Minnesota, June. Association for Computational Linguistics. [↗](https://doi.org/10.18653/v1/n19-1423)
11. Batoul Haidar, Maroun Chamoun, and Ahmed Serhrouchni. 2017. Multilingual cyberbullying detection system: Detecting cyberbullying in arabic content. In 2017 1st Cyber Security in Networking Conference (CSNet), pages 1–8, October. [find](https://scholar.google.com/scholar?q=Batoul+Haidar%2C+Maroun+Chamoun%2C+and+Ahmed+Serhrouchni%2E+2017%2E+Multilingual+cyberbullying+detection+system%3A+Detecting+cyberbullying+in+arabic+content%2E+In+2017+1st+Cyber+Security+in+Networking+Conference+)
12. Batoul Haidar, Maroun Chamoun, and Ahmed Serhrouchni. 2018. Arabic cyberbullying detection: Using deep learning. In 2018 7th International Conference on Computer and Communication Engineering (ICCCE), pages 284–289, September. [find](https://scholar.google.com/scholar?q=Batoul+Haidar%2C+Maroun+Chamoun%2C+and+Ahmed+Serhrouchni%2E+2018%2E+Arabic+cyberbullying+detection%3A+Using+deep+learning%2E+In+2018+7th+International+Conference+on+Computer+and+Communication+Engineering+%28ICCCE%29%2C)
13. Batoul Haidar, Maroun Chamoun, and Ahmed Serhrouchni. 2019. Arabic cyberbullying detection: Enhancing performance by using ensemble machine learning. In 2019 International Conference on Internet of Things (iThings) and IEEE Green Computing and Communications (GreenCom) and IEEE Cyber, Physical and Social Computing (CPSCom) and IEEE Smart Data (SmartData), pages 323–327, July. [find](https://scholar.google.com/scholar?q=Batoul+Haidar%2C+Maroun+Chamoun%2C+and+Ahmed+Serhrouchni%2E+2019%2E+Arabic+cyberbullying+detection%3A+Enhancing+performance+by+using+ensemble+machine+learning%2E+In+2019+International+Conference+on+Internet+of+Th)
14. 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 The 4th Workshop on Open-Source Arabic Corpora and Processing Tools (OSACT4). Proceedings of the Twelfth International Conference on Language Resources and Evaluation (LREC 2020), Marseille, France, May. [↗](https://doi.org/10.48550/arxiv.2005.07297)
15. Zachary Keyser. 2018. Gab, the ’alt-right twitter’ used by antisemitic pittsburgh shooter. The Jerusalem Post, October. [find](https://scholar.google.com/scholar?q=Zachary+Keyser%2E+2018%2E+Gab%2C+the+%E2%80%99alt%2Dright+twitter%E2%80%99+used+by+antisemitic+pittsburgh+shooter%2E+The+Jerusalem+Post%2C+October%2E)
16. Hanane Mohaouchane, Asmaa Mourhir, and Nikola Nikolov. 2019. Detecting offensive language on arabic social media using deep learning. In 2019 Sixth International Conference on Social Networks Analysis, Management and Security (SNAMS), pages 466–471, October. [find](https://scholar.google.com/scholar?q=Hanane+Mohaouchane%2C+Asmaa+Mourhir%2C+and+Nikola+Nikolov%2E+2019%2E+Detecting+offensive+language+on+arabic+social+media+using+deep+learning%2E+In+2019+Sixth+International+Conference+on+Social+Networks+Analysis)
17. Hamdy Mubarak, Kareem Darwish, and Walid Magdy. 2017. [Abusive language detection on Arabic social media](https://md.research.elamur.ai/free/en/papers/mubarak-2017-abusive-language-detection-arabic-social/). In Proceedings of the First Workshop on Abusive Language Online, pages 52–56, Vancouver, BC, Canada, August. Association for Computational Linguistics.
18. Hamdy Mubarak, Ammar Rashed, Kareem Darwish, Younes Samih, and Ahmed Abdelali. 2020. [Arabic offensive language on twitter: Analysis and experiments](https://md.research.elamur.ai/free/en/papers/mubarak-2021-arabic-offensive-language-twitter-analysis/). arXiv preprint arXiv:2004.02192. [↗](https://arxiv.org/abs/2004.02192)
19. Maarten Sap, Dallas Card, Saadia Gabriel, Yejin Choi, and Noah Smith. 2019. [The risk of racial bias in hate speech detection](https://md.research.elamur.ai/free/en/papers/sap-2019-risk-racial-bias-hate-speech/). In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, page 1668–1678. Association for Computational Linguistics.
20. Abu Bakr Soliman, Kareem Eissa, and Samhaa R. El-Beltagy. 2017. Aravec: A set of arabic word embedding models for use in arabic nlp. Procedia Computer Science, 117:256 – 265. Arabic Computational Linguistics. [find](https://scholar.google.com/scholar?q=Abu+Bakr+Soliman%2C+Kareem+Eissa%2C+and+Samhaa+R%2E+El%2DBeltagy%2E+2017%2E+Aravec%3A+A+set+of+arabic+word+embedding+models+for+use+in+arabic+nlp%2E+Procedia+Computer+Science%2C+117%3A256+%E2%80%93+265%2E+Arabic+Computational+Ling)
21. Marcos Zampieri, Preslav Nakov, Sara Rosenthal, Pepa Atanasova, Georgi Karadzhov, Hamdy Mubarak, Leon Derczynski, Zeses Pitenis, and Çağrı Çöltekin. 2020. SemEval-2020 Task 12: Multilingual Offensive Language Identification in Social Media (OffensEval 2020). In Proceedings of SemEval. [↗](https://doi.org/10.48550/arxiv.2006.07235)

*Updated 1 November 2026*