---
title: "Arabizi sentiment analysis based on transliteration and automatic"
description: "The paper tackles sentiment classification of Arabizi, Arabic (here mostly Algerian dialect) typed in… 25 references, cited by 9 papers here."
authors: ["Imane Guellil", "Ahsan Adeel", "Faical Azouaou", "Fodil Benali", "Ala-eddine Hachani", "Amir Hussain"]
year: 2018
venue: "Proceedings of the 9th Workshop on Computational Approaches to Subjectivity, Sentiment and Social Media Analysis"
source: "https://aclanthology.org/W18-6249/"
doi: "10.18653/v1/w18-6249"
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/guellil-2018-arabizi-sentiment-analysis-based-transliteration/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/guellil-2018-arabizi-sentiment-analysis-based-transliteration/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2018](https://md.research.elamur.ai/free/en/papers/2018/page/6/) / Arabizi sentiment analysis based on transliteration and automatic corpus annotation

# Arabizi sentiment analysis based on transliteration and automatic corpus annotation

**Authors:** [Imane Guellil](https://md.research.elamur.ai/free/en/authors/imane-guellil/), [Ahsan Adeel](https://md.research.elamur.ai/free/en/authors/ahsan-adeel/), [Faical Azouaou](https://md.research.elamur.ai/free/en/authors/faical-azouaou/), [Fodil Benali](https://md.research.elamur.ai/free/en/authors/fodil-benali/), [Ala-eddine Hachani](https://md.research.elamur.ai/free/en/authors/ala-eddine-hachani/) and [Amir Hussain](https://md.research.elamur.ai/free/en/authors/amir-hussain/) · **Year:** 2018 · **Venue:** [Proceedings of the 9th Workshop on Computational Approaches to Subjectivity, Sentiment and Social Media Analysis](https://md.research.elamur.ai/free/en/venues/wassa/) ([Workshop on Computational Approaches to Subjectivity, Sentiment and Social Media Analysis 2018](https://md.research.elamur.ai/free/en/venues/wassa/2018/)) · **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/W18-6249/) · [PDF](https://aclanthology.org/W18-6249.pdf) · [DOI 10.18653/v1/w18-6249](https://doi.org/10.18653/v1/w18-6249) · [OpenAlex](https://openalex.org/W2899793821)

**Citations:** 26 citations (OpenAlex, 2026-10-14) · [9](https://md.research.elamur.ai/free/en/papers/guellil-2018-arabizi-sentiment-analysis-based-transliteration/cited-by/) in our library, of the 149,439 papers whose bibliographies we read

## Summary

The paper tackles sentiment classification of Arabizi, Arabic (here mostly Algerian dialect) typed in Latin letters, digits and punctuation. Arabizi spelling varies wildly, and labelled data is scarce. The authors avoid manual labelling and parallel corpora with a pipeline of four steps:
1. They machine-translate the English SOCAL lexicon into Arabic and dialect through the Glosbe API, then clean it by hand.
2. They use that lexicon to score and label a large Facebook corpus written in Arabic script.
3. They convert Arabizi test messages into Arabic script using hand-written character rules. A frequency-based language model, built from a separate large Facebook corpus, picks the best candidate.
4. They train shallow classifiers on Doc2vec document vectors.

On a 500-message Algerian Arabizi test set, Naive Bayes with PV_DBOW reaches F1 of 0.78 when the test messages are transliterated by hand and 0.76 when transliterated automatically. This clearly beats an earlier 66% result that skipped transliteration. The work shows that a low-resource romanized dialect can be handled by mapping it to a better-resourced script and bootstrapping labels from a translated lexicon.

## Key points

- **Lexicon:** translating SOCAL gave 6,769 candidate terms. Manual review cut this to 1,745 Algerian-dialect entries: 968 negative, 6 neutral and 771 positive.
- **Training corpus:** the automatically labelled set is balanced, with 255,008 messages split into 127,504 positive and 127,504 negative.
- **Labelling rules:** a message's label is the average of its word scores. Scoring also handles negation (often an attached affix in dialect), multi-word entries, opposition, and a simple rule-based light stemmer.
- **Facebook data via RestFB:**
  - The September 2017 crawl has 8,673,285 messages, about 3.67M in Arabic script, and serves the transliteration language model.
  - The November 2017 crawl has 15,407,910 messages, about 7.93M in Arabic script, and serves sentiment annotation.
  - Corpus_50, drawn from Cotterell's corpus, tests transliteration; Corpus_500 (250 positive, 250 negative) tests sentiment.
- **Models compared:** three Doc2vec variants (PV-DM, PV-DBOW, and both combined) crossed with five classifiers (SVM, NB, LR, DT, RF). The best setup, PV_DBOW with NB, also has the top precision of 0.74 (manual) and 0.73 (automatic). Decision trees were consistently weakest.

## Interesting ideas
- Replacing a hand-built parallel corpus with rule-generated spelling candidates ranked by frequency in a large monolingual corpus is a cheap route to script normalisation.
- The authors find that ranking candidates by frequency alone selects common but wrong words, and they propose adding a distance measure between input and candidate.
- Projecting an English scored lexicon into another language by giving each translation its source score, averaging conflicts and then hand-reviewing, builds a lexicon quickly with a human check at the end.
- Testing the same classifier on both hand-transliterated and machine-transliterated inputs isolates how much error the upstream module adds. Here that cost was roughly two F1 points.
- Error analysis traced some misclassifications to wrong silver labels from the lexicon-based annotator, so noise in automatic training labels shows up directly in test errors.

## For Mova

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

- **I1** (tokens): noisy English input (texting spellings, dropped vowels) could get a rule-based candidate generator plus corpus-frequency ranking, and the chosen candidate and its rule would be logged for explainability.
- **I2** (tokens): any spelling-repair step should score candidates on corpus frequency plus edit distance, so a frequent word never overrides a near-exact match.
- **I3** (world): the same project-average-review loop could carry valence or measure attributes from one compiled resource onto WordNet/Wikidata entries that lack them, with conflicts flagged for review.
- **I4** (learn): evaluate downstream stages on gold-tree versus predicted-tree input so each stage's share of the error is measured explicitly.
- **I5** (learn): silver data agreed by witness parsers should get spot-checked samples, with label-noise errors counted separately from model errors.

*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

- [University of Stirling](https://md.research.elamur.ai/free/en/universities/university-of-stirling/) — University, Stirling, United Kingdom · 5 works · 2,653 citations
- [École Nationale Supérieure d'Informatique](https://md.research.elamur.ai/free/en/universities/ecole-nationale-superieure-d-informatique/) — University, Algiers, Algeria · 8 works · 430 citations

📚 Shelves (6): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 18,614 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 17,965 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 17,450 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 17,296 works · [💭 Psychology](https://md.research.elamur.ai/free/en/shelves/psychology/) · 1,649 works · [💭 Experimental and Cognitive Psychology](https://md.research.elamur.ai/free/en/shelves/psychology/experimental-and-cognitive-psychology/) · 930 works

🧭 Topics (2): [😊 Sentiment & emotion](https://md.research.elamur.ai/free/en/topics/sentiment-emotion/) · 1,035 works · [Semitic languages](https://md.research.elamur.ai/free/en/topics/semitic-languages/) · 3,066 works

## Cite

Cite as: Imane Guellil, Ahsan Adeel, Faical Azouaou, Fodil Benali, Ala-eddine Hachani and Amir Hussain. 2018. Arabizi sentiment analysis based on transliteration and automatic corpus annotation. Proceedings of the 9th Workshop on Computational Approaches to Subjectivity, Sentiment and Social Media Analysis.

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

```bibtex
@inproceedings{guellil2018arabizi,
  title = {{Arabizi sentiment analysis based on transliteration and automatic corpus annotation}},
  author = {Imane Guellil, Ahsan Adeel, Faical Azouaou, Fodil Benali, Ala-eddine Hachani and Amir Hussain},
  booktitle = {Proceedings of the 9th Workshop on Computational Approaches to Subjectivity, Sentiment and Social Media Analysis},
  year = {2018},
  doi = {10.18653/v1/w18-6249},
  url = {https://doi.org/10.18653/v1/w18-6249}
}
```

- [← previous QUD-Based Annotation of Discourse Structure and Information Structure: Tool and Evaluation](https://md.research.elamur.ai/free/en/papers/kuthy-2018-qud-based-annotation-discourse-structure/)
- [next → Co-speech gesture projection: Evidence from inferential judgments](https://md.research.elamur.ai/free/en/papers/tieu-2018-co-speech-gesture-projection-evidence/)
- [↑ back to papers of 2018](https://md.research.elamur.ai/free/en/papers/2018/page/6/)

## References

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

1. Muhammad Abdul-Mageed and Mona T Diab. 2012. [Awatif: A multi-genre corpus for modern standard arabic subjectivity and sentiment analysis](https://md.research.elamur.ai/free/en/papers/mageed-2012-awatif-multi-genre-corpus-modern/). In LREC, pages 3907–3914. Citeseer.
2. Nawaf A Abdulla, Nizar A Ahmed, Mohammed A Shehab, and Mahmoud Al-Ayyoub. 2013. Arabic sentiment analysis: Lexicon-based and corpus-based. In Applied Electrical Engineering and Computing Technologies (AEECT), 2013 IEEE Jordan Conference on, pages 1–6. IEEE. [find](https://scholar.google.com/scholar?q=Nawaf+A+Abdulla%2C+Nizar+A+Ahmed%2C+Mohammed+A+Shehab%2C+and+Mahmoud+Al%2DAyyoub%2E+2013%2E+Arabic+sentiment+analysis%3A+Lexicon%2Dbased+and+corpus%2Dbased%2E+In+Applied+Electrical+Engineering+and+Computing+Technologies+)
3. Imane Guellil, Faiçal Azouaou, and Mourad Abbas. 2017a. Comparison between neural and sta-tistical translation after translitera-tion of algerian arabic dialect. In WiNLP: Women & Underrepresented Minorities in Natural Language Processing (co-located withACL 2017). [find](https://scholar.google.com/scholar?q=Imane+Guellil%2C+Fai%C3%A7al+Azouaou%2C+and+Mourad+Abbas%2E+2017a%2E+Comparison+between+neural+and+sta%2Dtistical+translation+after+translitera%2Dtion+of+algerian+arabic+dialect%2E+In+WiNLP%3A+Women+%26+Underrepresented+Min)
4. Abdulaziz M Alayba, Vasile Palade, Matthew England, and Rahat Iqbal. 2017. [Arabic language sentiment analysis on health services](https://md.research.elamur.ai/free/en/papers/alayba-2017-arabic-language-sentiment-analysis-health/). In Arabic Script Analysis and Recognition (ASAR), 2017 1st International Workshop on, pages 114–118. IEEE.
5. Imane Guellil, Faical Azouaou, and Mourad Abbas. 2017b. Neural vs statistical translation of algerian arabic dialect written with arabizi and arabic letter. In The 31st Pacific Asia Conference on Language, Information and Computation PACLIC 31 (2017). [find](https://scholar.google.com/scholar?q=Imane+Guellil%2C+Faical+Azouaou%2C+and+Mourad+Abbas%2E+2017b%2E+Neural+vs+statistical+translation+of+algerian+arabic+dialect+written+with+arabizi+and+arabic+letter%2E+In+The+31st+Pacific+Asia+Conference+on+Lang)
6. Mohamed Aly and Amir Atiya. 2013. [Labr: A large scale arabic book reviews dataset](https://md.research.elamur.ai/free/en/papers/aly-2013-labr-large-scale-arabic-book/). In Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), volume 2, pages 494–498. [↗](https://doi.org/10.13140/2.1.3960.5761)
7. Imane Guellil, Faiçal Azouaou, Mourad Abbas, and Sadat Fatiha. 2017c. Arabizi transliteration of algerian arabic dialect into modern standard arabic. In Social MT 2017/First workshop on Social Media and User Generated Content Machine Translation. [find](https://scholar.google.com/scholar?q=Imane+Guellil%2C+Fai%C3%A7al+Azouaou%2C+Mourad+Abbas%2C+and+Sadat+Fatiha%2E+2017c%2E+Arabizi+transliteration+of+algerian+arabic+dialect+into+modern+standard+arabic%2E+In+Social+MT+2017%2FFirst+workshop+on+Social+Media+a)
8. Faical Azouaou and Imane Guellil. 2017. Alg/fr: A step by step construction of a lexicon between algerian dialect and french. In The 31st Pacific Asia Conference on Language, Information and Computation PACLIC 31 (2017). [find](https://scholar.google.com/scholar?q=Faical+Azouaou+and+Imane+Guellil%2E+2017%2E+Alg%2Ffr%3A+A+step+by+step+construction+of+a+lexicon+between+algerian+dialect+and+french%2E+In+The+31st+Pacific+Asia+Conference+on+Language%2C+Information+and+Computati)
9. Imane Guellil and Azouaou Faical. 2017. Bilingual lexicon for algerian arabic dialect treatment in social media. In WiNLP: Women & Underrepresented Minorities in Natural Language Processing (co-located with ACL 2017). [http://www.winlp.org/wp-content/uploads/2017/final_papers_2017/92_Paper.pdf](http://www.winlp.org/wp-content/uploads/2017/final_papers_2017/92_Paper.pdf). [↗](http://www.winlp.org/wp-content/uploads/2017/final_papers_2017/92_Paper.pdf)
10. 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.
11. Imene Guellil and Kamel Boukhalfa. 2015. Social big data mining: A survey focused on opinion mining and sentiments analysis. In Programming and Systems (ISPS), 2015 12th International Symposium on, pages 1–10. IEEE. [find](https://scholar.google.com/scholar?q=Imene+Guellil+and+Kamel+Boukhalfa%2E+2015%2E+Social+big+data+mining%3A+A+survey+focused+on+opinion+mining+and+sentiments+analysis%2E+In+Programming+and+Systems+%28ISPS%29%2C+2015+12th+International+Symposium+on%2C+pa)
12. Ann Bies, Zhiyi Song, Mohamed Maamouri, Stephen Grimes, Haejoong Lee, Jonathan Wright, Stephanie Strassel, Nizar Habash, Ramy Eskander, and Owen Rambow. 2014. [Transliteration of arabizi into arabic orthography: Developing a parallel annotated arabizi-arabic script sms/chat corpus](https://md.research.elamur.ai/free/en/papers/bies-2014-transliteration-arabizi-into-arabic-orthography/). In Proceedings of the EMNLP 2014 Workshop on Arabic Natural Language Processing (ANLP), pages 93–103.
13. Kamaljeet Kaur and Parminder Singh. 2014. Review of machine transliteration techniques. International Journal of Computer Applications, 107(20). [↗](https://doi.org/10.5120/18866-0061)
14. Aamera ZH Khan, Mohammad Atique, and VM Thakare. 2015. Combining lexicon-based and learning-based methods for twitter sentiment analysis. International Journal of Electronics, Communication and Soft Computing Science & Engineering (IJECSCSE), page 89. [find](https://scholar.google.com/scholar?q=Aamera+ZH+Khan%2C+Mohammad+Atique%2C+and+VM+Thakare%2E+2015%2E+Combining+lexicon%2Dbased+and+learning%2Dbased+methods+for+twitter+sentiment+analysis%2E+International+Journal+of+Electronics%2C+Communication+and+Soft+C)
15. Ryan Cotterell and Chris Callison-Burch. 2014. [A multi-dialect, multi-genre corpus of informal written arabic](https://md.research.elamur.ai/free/en/papers/cotterell-2014-multi-dialect-multi-genre-corpus/).
16. Kareem Darwish. 2013. [Arabizi detection and conversion to arabic](https://md.research.elamur.ai/free/en/papers/darwish-2014-arabizi-detection-conversion-arabic/). arXiv preprint arXiv:1306.6755. [↗](https://arxiv.org/abs/1306.6755)
17. Quoc Le and Tomas Mikolov. 2014. [Distributed representations of sentences and documents](https://md.research.elamur.ai/free/en/papers/le-2014-distributed-representations-sentences-documents/). In International Conference on Machine Learning, pages 1188–1196. [↗](https://doi.org/10.48550/arxiv.1405.4053)
18. Bing Liu. 2012. Sentiment analysis and opinion mining. Synthesis lectures on human language technologies, 5(1):1–167. [↗](https://doi.org/10.1007/978-3-031-02145-9)
19. Andrew L Maas, Raymond E Daly, Peter T Pham, Dan Huang, Andrew Y Ng, and Christopher Potts. 2011. [Learning word vectors for sentiment analysis](https://md.research.elamur.ai/free/en/papers/maas-2011-learning-word-vectors-sentiment-analysis/). In Proceedings of the 49th annual meeting of the association for computational linguistics: Human language technologies-volume 1, pages 142–150. Association for Computational Linguistics.
20. Karima Meftouh, Najette Bouchemal, and Kamel Smaïli. 2012. A study of a non-resourced language: The case of one of the algerian dialects. In The third International Workshop on Spoken Languages Technologies for Under-resourced Languages-SLTU’12. [find](https://scholar.google.com/scholar?q=Karima+Meftouh%2C+Najette+Bouchemal%2C+and+Kamel+Sma%C3%AFli%2E+2012%2E+A+study+of+a+non%2Dresourced+language%3A+The+case+of+one+of+the+algerian+dialects%2E+In+The+third+International+Workshop+on+Spoken+Languages+Techno)
21. Mahmoud Nabil, Mohamed Aly, and Amir Atiya. 2015. [Astd: Arabic sentiment tweets dataset](https://md.research.elamur.ai/free/en/papers/nabil-2015-astd-arabic-sentiment-tweets-dataset/). In Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, pages 2515–2519. Houda Saâdane1 Damien Nouvel, Hosni Seffih, and Christian Fluhr. Une approche linguistique pour la détection des dialectes arabes. In 24e Conférence sur le Traitement Automatique des Langues Naturelles (TALN), page 242. [↗](https://doi.org/10.60692/3ee2g-zjt73)
22. Cotterell Ryan, Adithya Renduchintala, Naomi Saphra, and Chris Callison-Burch. 2014. An algerian arabic-french code-switched corpus. In Workshop on Free/Open-Source Arabic Corpora and Corpora Processing Tools Workshop Programme, page 34. [find](https://scholar.google.com/scholar?q=Cotterell+Ryan%2C+Adithya+Renduchintala%2C+Naomi+Saphra%2C+and+Chris+Callison%2DBurch%2E+2014%2E+An+algerian+arabic%2Dfrench+code%2Dswitched+corpus%2E+In+Workshop+on+Free%2FOpen%2DSource+Arabic+Corpora+and+Corpora+Processi)
23. Maite Taboada, Julian Brooke, Milan Tofiloski, Kimberly Voll, and Manfred Stede. 2011. [Lexicon-based methods for sentiment analysis](https://md.research.elamur.ai/free/en/papers/taboada-2011-lexicon-based-methods-sentiment-analysis/). Computational linguistics, 37(2):267–307. [↗](https://doi.org/10.1162/coli_a_00049)
24. Mike Thelwall, Kevan Buckley, Georgios Paltoglou, Di Cai, and Arvid Kappas. 2010. Sentiment strength detection in short informal text. Journal of the American Society for Information Science and Technology, 61(12):2544–2558. [↗](https://doi.org/10.1002/asi.21416)
25. Marlies van der Wees, Arianna Bisazza, and Christof Monz. 2016. [A simple but effective approach to improve arabizi-to-english statistical machine translation](https://md.research.elamur.ai/free/en/papers/wees-2016-simple-but-effective-approach-improve/). In Proceedings of the 2nd Workshop on Noisy User-generated Text (WNUT), pages 43–50.

*Updated 2 November 2026*