---
title: "Subjective Sentiment Analysis for Arabic — Bessou 2019"
description: "The paper addresses polarity classification of reader comments on Arabic news sites, labeling each as positive, negative or neutral."
authors: ["Sadik Bessou", "Rania Aberkane"]
year: 2019
venue: "arXiv:1911.03776 · Journal of Digital Information Management ISSN 0972-7272 2019"
source: "https://arxiv.org/abs/1911.03776"
doi: "10.6025/jdim/2019/17/5/289-295"
arxiv: "1911.03776"
licence: "arXiv — non-exclusive license to distribute"
licence_url: "http://arxiv.org/licenses/nonexclusive-distrib/1.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/bessou-2019-subjective-sentiment-analysis-arabic-newswire/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/bessou-2019-subjective-sentiment-analysis-arabic-newswire/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2019](https://md.research.elamur.ai/free/en/papers/2019/page/4/) / Subjective Sentiment Analysis for Arabic Newswire Comments

# Subjective Sentiment Analysis for Arabic Newswire Comments

**Authors:** [Sadik Bessou](https://md.research.elamur.ai/free/en/authors/sadik-bessou/), Rania Aberkane · **Year:** 2019 · **Venue:** arXiv:1911.03776 · Journal of Digital Information Management ISSN 0972-7272 2019 · **Licence:** [arXiv — non-exclusive license to distribute](http://arxiv.org/licenses/nonexclusive-distrib/1.0/)

**Links:** [Publisher's page](https://doi.org/10.6025/jdim/2019/17/5/289-295) · [PDF](https://arxiv.org/pdf/1911.03776) · [DOI 10.6025/jdim/2019/17/5/289-295](https://doi.org/10.6025/jdim/2019/17/5/289-295) · [arXiv 1911.03776](https://arxiv.org/abs/1911.03776) · [OpenAlex](https://openalex.org/W2985978206)

**Citations:** 5 citations (openalex-snapshot · 2026-10-14) · 1 in our library, of the 140884 papers whose bibliographies we read

## Summary
The paper addresses polarity classification of reader comments on Arabic news sites, labeling each as positive, negative or neutral. No shared benchmark exists for this setting, and users often mix MSA with dialect. The authors build a new sentence-level resource from Echorouk online, an Algerian daily. They gathered comments by hand from articles on politics, economy, culture and similar topics, kept only articles with more than 20 opinionated comments, and labeled every comment manually.

A cleaning pipeline removes links and non-Arabic noise, unifies Hamza/Alif forms, shortens elongated words and drops stop words. The comments are then encoded as count or TF-IDF vectors over word uni-grams, bi-grams or both. Six classifiers are compared: Multinomial Naïve Bayes, linear SVM, Random Forest, Logistic Regression, MLP and KNN.

Multinomial Naïve Bayes on count vectors with combined uni- and bi-grams performs best. It reaches 85.57% accuracy for positive vs negative and 65.64% once neutral is added. The contribution is mainly a dialect-tinged newswire-comment dataset plus a clear baseline showing that cheap lexical features still carry much of the signal.

## Key points
- Corpus: 1,633 comments and 63,055 tokens from Echorouk online. Tokens per label are 31,392 negative, 21,248 neutral and 9,975 positive, so the classes are strongly skewed.
- Representations and features: count vectors and TF-IDF over word uni-grams, bi-grams, and the two combined.
- Two-class accuracy for MNB is 84.71% with uni-grams, 75.20% with bi-grams only and 85.57% with both. MLP is second at 83.05%, and KNN trails at 40.49%.
- Three-class accuracy for MNB is 64.41% with uni-grams, 58.58% with bi-grams (tied with LR) and 65.64% with both. KNN drops to 31.90%.
- Bi-grams alone underperform, and adding uni-grams gains roughly 10 points (two-class) and 7 points (three-class). Including the neutral class costs about 20 points of accuracy.

## Interesting ideas
- Collapsing letters that users repeat for emphasis into a single occurrence shrinks the vocabulary without losing the word, and the fact that it was elongated can be kept as a separate signal of intensity. — For Mova (tokens): The en tokenizer could map forms like "sooooo"/"goooood" to their lemma while attaching an emphasis feature, so the tagger sees a known word and pragmatics still knows it was stressed.
- Harvesting only articles that attracted more than 20 comments is a cheap filter that concentrates opinionated, argumentative text in the collected data. — For Mova (learn): When gathering raw text for silver data on evaluative or argumentative language, a similar engagement threshold could select documents dense in stance and speech acts rather than neutral prose.
- Reporting the same systems on both a two-way and a three-way split exposes how much difficulty the neutral class adds, which a single score would hide. — For Mova (answer): Evaluations of attitude or stance questions could be run with and without a "no opinion / neutral" option, revealing whether the reader truly detects absence of stance or just picks a pole.
- A plain Multinomial Naïve Bayes on word n-gram counts beat MLP and other heavier models here, making it a strong and inexpensive reference point. — For Mova (learn): Such a bag-of-n-grams classifier is worth keeping as a negative control, so any explainable tree-and-frame component must show it beats what surface lexical statistics already achieve.
- Bi-grams alone were too sparse, but combined with uni-grams they added useful local context. Wider units help most when backed off to smaller ones. — For Mova (frames): Polarity cues in prag could combine single-word valence with valence attached to dependency pairs (e.g. negation + head, modifier + noun), falling back to the word level when the pair is unseen.

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

Fields: [Linguistics & languages](https://md.research.elamur.ai/free/en/shelves/linguistics/)

## Cite

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

```bibtex
@misc{bessou2019subjective,
  title = {{Subjective Sentiment Analysis for Arabic Newswire Comments}},
  author = {Bessou, Sadik and Aberkane, Rania},
  year = {2019},
  doi = {10.6025/jdim/2019/17/5/289-295},
  eprint = {1911.03776},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.6025/jdim/2019/17/5/289-295}
}
```

- [← previous Style Transformer: Unpaired Text Style Transfer without Disentangled Latent Representation](https://md.research.elamur.ai/free/en/papers/dai-2019-style-transformer-unpaired-text-style/)
- [next → Submodular Optimization-based Diverse Paraphrasing and its Effectiveness in Data Augmentation](https://md.research.elamur.ai/free/en/papers/kumar-2019-submodular-optimization-based-diverse-paraphrasing/)
- [↑ back to papers of 2019](https://md.research.elamur.ai/free/en/papers/2019/page/4/)

## References

23 references, 4 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Abdul-Mageed, M. (2018). Learning Subjective Language: Feature Engineered vs. Deep Models. In: LRECWorkshop The 3rd Workshop on Open-Source Arabic Corpora and Processing Tools. 80-90. [find](https://scholar.google.com/scholar?q=Abdul%2DMageed%2C+M%2E+%282018%29%2E+Learning+Subjective+Language%3A+Feature+Engineered+vs%2E+Deep+Models%2E+In%3A+LRECWorkshop+The+3rd+Workshop+on+Open%2DSource+Arabic+Corpora+and+Processing+Tools%2E+80%2D90%2E)
2. Abdul-Mageed, M., Diab, M. T. (2012). AWATIF: A Multi-Genre Corpus for Modern Standard Arabic Subjectivity and Sentiment Analysis. In: LREC, 3907-3914. [find](https://scholar.google.com/scholar?q=Abdul%2DMageed%2C+M%2E%2C+Diab%2C+M%2E+T%2E+%282012%29%2E+AWATIF%3A+A+Multi%2DGenre+Corpus+for+Modern+Standard+Arabic+Subjectivity+and+Sentiment+Analysis%2E+In%3A+LREC%2C+3907%2D3914%2E)
3. Abdul-Mageed, M., Diab, M. T. (2014). SANA: A Large Scale Multi-Genre, Multi-Dialect Lexicon for Arabic Subjectivity and Sentiment Analysis. In: LREC, 1162-1169. [find](https://scholar.google.com/scholar?q=Abdul%2DMageed%2C+M%2E%2C+Diab%2C+M%2E+T%2E+%282014%29%2E+SANA%3A+A+Large+Scale+Multi%2DGenre%2C+Multi%2DDialect+Lexicon+for+Arabic+Subjectivity+and+Sentiment+Analysis%2E+In%3A+LREC%2C+1162%2D1169%2E)
4. Al-Ayyoub, M., Khamaiseh, A. A., Jararweh, Y., Al-Kabi, M. N. (2019). A comprehensive survey of arabic sentiment analysis. Information Processing & Management, 56(2) 320-342. [find](https://scholar.google.com/scholar?q=Al%2DAyyoub%2C+M%2E%2C+Khamaiseh%2C+A%2E+A%2E%2C+Jararweh%2C+Y%2E%2C+Al%2DKabi%2C+M%2E+N%2E+%282019%29%2E+A+comprehensive+survey+of+arabic+sentiment+analysis%2E+Information+Processing+%26+Management%2C+56%282%29+320%2D342%2E)
5. Alomari, K. M., ElSherif, H. M., Shaalan, K. (2017). Arabic Tweets Sentimental Analysis Using Machine Learning. In: International Conference on Industrial, Engineering and Other Applications of Applied Intelligent Systems. 602-610. Springer, Cham. [find](https://scholar.google.com/scholar?q=Alomari%2C+K%2E+M%2E%2C+ElSherif%2C+H%2E+M%2E%2C+Shaalan%2C+K%2E+%282017%29%2E+Arabic+Tweets+Sentimental+Analysis+Using+Machine+Learning%2E+In%3A+International+Conference+on+Industrial%2C+Engineering+and+Other+Applications+of+Applie)
6. Aly, M., Atiya, A. (2013). LABR: A large scale Arabic book reviews dataset. In: Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers). 2, 494-498. [find](https://scholar.google.com/scholar?q=Aly%2C+M%2E%2C+Atiya%2C+A%2E+%282013%29%2E+LABR%3A+A+large+scale+Arabic+book+reviews+dataset%2E+In%3A+Proceedings+of+the+51st+Annual+Meeting+of+the+Association+for+Computational+Linguistics+%28Volume+2%3A+Short+Papers%29%2E+2%2C+494)
7. Assiri, A., Emam, A., Aldossari, H. (2015). Arabic sentiment analysis: a survey. International Journal of Advanced Computer Science and Applications, 6(12) 75-85. [find](https://scholar.google.com/scholar?q=Assiri%2C+A%2E%2C+Emam%2C+A%2E%2C+Aldossari%2C+H%2E+%282015%29%2E+Arabic+sentiment+analysis%3A+a+survey%2E+International+Journal+of+Advanced+Computer+Science+and+Applications%2C+6%2812%29+75%2D85%2E)
8. Baly, R., Khaddaj, A., Hajj, H., El-Hajj, W., Shaban, K. B. (2019). ArSentD-LEV: A multi-topic corpus for target-based sentiment analysis in Arabic levantine tweets. arXiv preprint arXiv:1906.01830. [↗](https://arxiv.org/abs/1906.01830) [our card](https://md.research.elamur.ai/free/en/papers/baly-2019-arsentd-lev-multi-topic-corpus/)
9. Bessou, S., Touahria, M. (2014). An Accuracy-Enhanced Stemming Algorithm for Arabic Information Retrieval. Neural Network World, 24(2) 117-128. [find](https://scholar.google.com/scholar?q=Bessou%2C+S%2E%2C+Touahria%2C+M%2E+%282014%29%2E+An+Accuracy%2DEnhanced+Stemming+Algorithm+for+Arabic+Information+Retrieval%2E+Neural+Network+World%2C+24%282%29+117%2D128%2E) [our card](https://md.research.elamur.ai/free/en/papers/bessou-2019-accuracy-enhanced-stemming-algorithm-arabic/)
10. Boudad, N., Faizi, R., Thami, R. O. H., Chiheb, R. (2017). Sentiment analysis in Arabic: A review of the literature. Ain Shams Engineering Journal. [find](https://scholar.google.com/scholar?q=Boudad%2C+N%2E%2C+Faizi%2C+R%2E%2C+Thami%2C+R%2E+O%2E+H%2E%2C+Chiheb%2C+R%2E+%282017%29%2E+Sentiment+analysis+in+Arabic%3A+A+review+of+the+literature%2E+Ain+Shams+Engineering+Journal%2E)
11. Elarnaoty, M., AbdelRahman, S., Fahmy, A. (2012). A machine learning approach for opinion holder extraction in Arabic language. arXiv preprint arXiv:1206.1011. [↗](https://arxiv.org/abs/1206.1011)
12. El-Masri, M., Altrabsheh, N., Mansour, H., Ramsay, A. (2017). A web-based tool for Arabic sentiment analysis. Procedia Computer Science, 117, 38-45. [find](https://scholar.google.com/scholar?q=El%2DMasri%2C+M%2E%2C+Altrabsheh%2C+N%2E%2C+Mansour%2C+H%2E%2C+Ramsay%2C+A%2E+%282017%29%2E+A+web%2Dbased+tool+for+Arabic+sentiment+analysis%2E+Procedia+Computer+Science%2C+117%2C+38%2D45%2E)
13. ElSahar, H., El-Beltagy, S. R. (2015). Building large Arabic multi-domain resources for sentiment analysis. In: International Conference on Intelligent Text Processing and Computational Linguistics. 23-34. Springer, Cham. [find](https://scholar.google.com/scholar?q=ElSahar%2C+H%2E%2C+El%2DBeltagy%2C+S%2E+R%2E+%282015%29%2E+Building+large+Arabic+multi%2Ddomain+resources+for+sentiment+analysis%2E+In%3A+International+Conference+on+Intelligent+Text+Processing+and+Computational+Linguistics%2E+2)
14. Heikal, M., Torki, M., El-Makky, N. (2018). Sentiment Analysis of Arabic Tweets using Deep Learning. Procedia Computer Science, 142, 114-122. [find](https://scholar.google.com/scholar?q=Heikal%2C+M%2E%2C+Torki%2C+M%2E%2C+El%2DMakky%2C+N%2E+%282018%29%2E+Sentiment+Analysis+of+Arabic+Tweets+using+Deep+Learning%2E+Procedia+Computer+Science%2C+142%2C+114%2D122%2E)
15. Ibrahim, H. S., Abdou, S. M., Gheith, M. (2015). Sentiment analysis for modern standard Arabic and colloquial. arXiv preprint arXiv:1505.03105. [↗](https://arxiv.org/abs/1505.03105)
16. Khalifa, K., Omar, N. (2014). A Hybrid method using Lexicon-based Approach and Naive Bayes Classifier for Arabic Opinion Question Answering. JCS, 10(10) 1961-1968. [find](https://scholar.google.com/scholar?q=Khalifa%2C+K%2E%2C+Omar%2C+N%2E+%282014%29%2E+A+Hybrid+method+using+Lexicon%2Dbased+Approach+and+Naive+Bayes+Classifier+for+Arabic+Opinion+Question+Answering%2E+JCS%2C+10%2810%29+1961%2D1968%2E)
17. Khoja, S., Garside, R. (1999). Stemming Arabic text. Lancaster, UK, Computing Department, Lancaster University. [find](https://scholar.google.com/scholar?q=Khoja%2C+S%2E%2C+Garside%2C+R%2E+%281999%29%2E+Stemming+Arabic+text%2E+Lancaster%2C+UK%2C+Computing+Department%2C+Lancaster+University%2E)
18. Larkey, L. S., Connell, M. E. (2001). Arabic information retrieval at UMass in TREC-10. In: TREC. [find](https://scholar.google.com/scholar?q=Larkey%2C+L%2E+S%2E%2C+Connell%2C+M%2E+E%2E+%282001%29%2E+Arabic+information+retrieval+at+UMass+in+TREC%2D10%2E+In%3A+TREC%2E)
19. Mourad, A., Darwish, K. (2013). Subjectivity and sentiment analysis of modern standard Arabic and Arabic microblogs. In: Proceedings of the 4th workshop on computational approaches to subjectivity, sentiment and social media analysis (p. 55-64). [find](https://scholar.google.com/scholar?q=Mourad%2C+A%2E%2C+Darwish%2C+K%2E+%282013%29%2E+Subjectivity+and+sentiment+analysis+of+modern+standard+Arabic+and+Arabic+microblogs%2E+In%3A+Proceedings+of+the+4th+workshop+on+computational+approaches+to+subjectivity%2C+se)
20. Nabil, M., Aly, M., Atiya, A. (2015). ASTD: Arabic sentiment tweets dataset. In: Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, 2515-2519. [find](https://scholar.google.com/scholar?q=Nabil%2C+M%2E%2C+Aly%2C+M%2E%2C+Atiya%2C+A%2E+%282015%29%2E+ASTD%3A+Arabic+sentiment+tweets+dataset%2E+In%3A+Proceedings+of+the+2015+Conference+on+Empirical+Methods+in+Natural+Language+Processing%2C+2515%2D2519%2E) [our card](https://md.research.elamur.ai/free/en/papers/nabil-2015-astd-arabic-sentiment-tweets-dataset/)
21. Rosenthal, S., Farra, N., Nakov, P. (2017). SemEval-2017 task 4: Sentiment analysis in Twitter. In: Proceedings of the 11th International Workshop on Semantic Evaluation (SemEval-2017) 502-518. [find](https://scholar.google.com/scholar?q=Rosenthal%2C+S%2E%2C+Farra%2C+N%2E%2C+Nakov%2C+P%2E+%282017%29%2E+SemEval%2D2017+task+4%3A+Sentiment+analysis+in+Twitter%2E+In%3A+Proceedings+of+the+11th+International+Workshop+on+Semantic+Evaluation+%28SemEval%2D2017%29+502%2D518%2E) [our card](https://md.research.elamur.ai/free/en/papers/rosenthal-2017-semeval-2017-task-4-sentiment/)
22. Shoukry, A., Rafea, A. (2012). Sentence-level Arabic sentiment analysis. In: 2012 International Conference on Collaboration Technologies and Systems (CTS) (p. 546-550). IEEE. [find](https://scholar.google.com/scholar?q=Shoukry%2C+A%2E%2C+Rafea%2C+A%2E+%282012%29%2E+Sentence%2Dlevel+Arabic+sentiment+analysis%2E+In%3A+2012+International+Conference+on+Collaboration+Technologies+and+Systems+%28CTS%29+%28p%2E+546%2D550%29%2E+IEEE%2E)
23. Tartir, S., Abdul-Nabi, I. (2017). Semantic sentiment analysis in Arabic social media. Journal of King Saud University-Computer and Information Sciences, 29(2) 229-233. [find](https://scholar.google.com/scholar?q=Tartir%2C+S%2E%2C+Abdul%2DNabi%2C+I%2E+%282017%29%2E+Semantic+sentiment+analysis+in+Arabic+social+media%2E+Journal+of+King+Saud+University%2DComputer+and+Information+Sciences%2C+29%282%29+229%2D233%2E)