Elamur Research⌕
☰
home / papers / 2023 / Content-Localization based System for Analyzing Sentiment and Hate Behaviors in Low-Resource Dialectal Arabic: English to Levantine and Gulf

Content-Localization based System for Analyzing Sentiment and Hate Behaviors in Low-Resource Dialectal Arabic: English to Levantine and Gulf

Summary

Social media posts in Arabic are mostly written in regional dialects, and labeled data for those dialects is scarce. Plain machine translation tends to lose the tone, idioms and online slang that carry sentiment or hostility. The authors therefore propose content localization: English sentiment and hate speech datasets are rendered into Levantine and Gulf Arabic by neural machine translation models adapted to informal social-media language and regional culture. The localized corpora are used to train four BERT classifiers, one for sentiment and one for hate speech in each dialect. The classifiers are tested in two ways: on separately collected, manually annotated native dialect data, and in a case study of 2020 COVID-19 posts from Lebanon and Saudi Arabia. To explain what drives the predictions, BERTopic discovers topics and RAKE extracts variable-length phrases that describe each topic. The classifiers perform strongly overall, but hate detection transfers unevenly between dialects. This shows that low-resource social behavior analysis can reuse English annotations, provided dialect differences are respected.

Key points

  • The source language is English and the targets are two low-resource Arabic dialects, Levantine and Gulf. The four resulting classifiers are BERT-based and cover sentiment and hate speech.
  • Evaluation does not rely on translated test sets. It uses externally collected, natively annotated Levantine and Gulf data, plus a case study of COVID-19 data from Lebanon and Saudi Arabia.
  • Dialect matters: the localized Levantine hate model detected native Levantine hate well, while the Gulf hate model performed poorly on its own native data.
  • BERTopic was used because it tolerates social-media noise without language-specific cleanup tools. RAKE then turns topic keywords into descriptive phrases whose length adapts to the topic rather than using a fixed window.
  • The related work notes that MADAR found only 25.8% average similarity between Arabic dialects. It also criticizes earlier English–dialect corpora for being domain-bound, small and blind to code switching, idioms and online slang.

Interesting ideas

  • I1. Instead of collecting new labeled data, the authors move an existing annotated corpus into the target variety while keeping its labels, adapting tone and idiom rather than words alone; this makes label transfer cheap. — For Mova (learn): Labeled English fable and frame data could be rewritten into informal registers (chat, slang) with labels preserved, giving silver training data for registers Mova rarely sees.
  • I2. The test data comes from native, independently annotated text rather than from the same localization pipeline that produced the training data, so pipeline artifacts cannot inflate scores. — For Mova (learn): Treat genuinely native, held-out text as a mandatory control whenever a model is trained on synthetic or converted data.
  • I3. Each variety gets its own model, and the comparison across varieties exposes where transfer silently breaks, as with the Gulf hate model. — For Mova (tree): When converting treebanks between UD dialects or registers, score the parser separately on native data of each variety to catch conversion gaps hidden by averaged metrics.
  • I4. Clusters are explained with phrases whose length adapts to the content, found by RAKE over topic keywords, instead of keyword lists or fixed n-gram windows. Phrases give context that bare keywords lack, while staying more general than full sentences. — For Mova (world): Entity or event clusters in the text world could be labeled with adaptive-length phrases from the source text, making the world summaries more readable and easier to explain.

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

Topics: Machine translation

Cited by

1 paper of our library cites it (of the 140884 whose bibliographies we read); each has a page here.

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.48550/arxiv.2312.03727 · arXiv 2312.03727

@misc{alzamzami2023content,
  title = {{Content-Localization based System for Analyzing Sentiment and Hate Behaviors in Low-Resource Dialectal Arabic: English to Levantine and Gulf}},
  author = {Alzamzami, Fatimah and Saddik, Abdulmotaleb El},
  year = {2023},
  doi = {10.48550/arxiv.2312.03727},
  eprint = {2312.03727},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2312.03727}
}

References

52 references, 11 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Vineet John and Olga Vechtomova. Sentiment analysis on financial news headlines using training dataset augmentation. arXiv preprint arXiv:1707.09448, 2017. ↗
  2. Hind Saleh Alatawi, Areej Maatog Alhothali, and Kawthar Mustafa Moria. Detecting white supremacist hate speech using domain specific word embedding with deep learning and bert. arXiv e-prints, pages arXiv–2010, 2020. find
  3. Meshrif Alruily and Osama R Shahin. Sentiment analysis of twitter data for saudi universities. International Journal of Machine Learning and Computing, 10(1), 2020. find our card
  4. Fatimah Alzamzami and Abdulmotaleb El Saddik. Monitoring cyber sentihate social behavior during covid-19 pandemic in north america. IEEE Access, 2021. find
  5. Mukesh Kumar Saini, Fatimah Al-Zamzami, and Abdulmotaleb El Saddik. Towards storytelling by extracting social information from osn photo’s metadata. In Proceedings of the First International Workshop on Internet-Scale Multimedia Management, pages 15–20, 2014. find
  6. Fatimah Alzamzami, Mukesh Saini, and Abdulmotaleb El Saddik. Dst: days spent together using soft sensory information on osns—a case study on facebook. Soft Computing, 21(15):4227–4238, 2017. find
  7. Fatimah Alzamzami, Mohamad Hoda, and Abdulmotaleb El Saddik. Light gradient boosting machine for general sentiment classification on short texts: a comparative evaluation. IEEE access, 8:101840–101858, 2020. find
  8. Abdulaziz M Alayba, Vasile Palade, Matthew England, and Rahat Iqbal. A combined cnn and lstm model for arabic sentiment analysis. In Machine Learning and Knowledge Extraction: Second IFIP TC 5, TC 8/WG 8.4, 8.9, TC 12/WG 12.9 International Cross-Domain Conference, CD-MAKE 2018, Hamburg, Germany, August 27–30, 2018, Proceedings 2, pages 179–191. Springer, 2018. find
  9. Hui Yin, Shuiqiao Yang, and Jianxin Li. Detecting topic and sentiment dynamics due to covid-19 pandemic using social media. arXiv preprint arXiv:2007.02304, 2020. ↗
  10. Raghad Alshalan, Hend Al-Khalifa, Duaa Alsaeed, Heyam Al-Baity, and Shahad Alshalan. Hate detection in covid-19 tweets in the arab region using deep learning and topic modeling. Journal of Medical Internet Research, 2020. find
  11. Qiaozhu Mei, Xuehua Shen, and ChengXiang Zhai. Automatic labeling of multinomial topic models. In Proceedings of the 13th ACM SIGKDD international conference on Knowledge discovery and data mining, pages 490–499, 2007. find
  12. Arwa Alqudsi, Nazlia Omar, and Khalid Shaker. A hybrid rules and statistical method for arabic to english machine translation. In 2019 2nd International Conference on Computer Applications & Information Security (ICCAIS), pages 1–7. IEEE, 2019. find
  13. Arwa Alqudsi, Nazlia Omar, and Khalid Shaker. Arabic machine translation: a survey. Artificial Intelligence Review, 42:549–572, 2014. find
  14. Jezia Zakraoui, Moutaz Saleh, Somaya Al-Maadeed, and Jihad Mohamad AlJa’am. Evaluation of arabic to english machine translation systems. In 2020 11th International Conference on Information and Communication Systems (ICICS), pages 185–190. IEEE, 2020. find
  15. Surangika Ranathunga, En-Shiun Annie Lee, Marjana Prifti Skenduli, Ravi Shekhar, Mehreen Alam, and Rishemjit Kaur. Neural machine translation for low-resource languages: A survey. ACM Computing Surveys, 55(11):1–37, 2023. find
  16. Hassan Sajjad, Ahmed Abdelali, Nadir Durrani, and Fahim Dalvi. Arabench: Benchmarking dialectal arabic-english machine translation. In Proceedings of the 28th International Conference on Computational Linguistics, pages 5094–5107, 2020. find
  17. Rabih Zbib, Erika Malchiodi, Jacob Devlin, David Stallard, Spyros Matsoukas, Richard Schwartz, John Makhoul, Omar Zaidan, and Chris Callison-Burch. Machine translation of arabic dialects. In Proceedings of the 2012 conference of the north american chapter of the association for computational linguistics: Human language technologies, pages 49–59, 2012. find our card
  18. El Moatez Billah Nagoudi, AbdelRahim Elmadany, and Muhammad Abdul-Mageed. Investigating code-mixed modern standard arabic-egyptian to english machine translation. arXiv preprint arXiv:2105.13573, 2021. ↗
  19. Houda Bouamor, Nizar Habash, and Kemal Oflazer. A multidialectal parallel corpus of arabic. In LREC, pages 1240–1245, 2014. find our card
  20. Houda Bouamor, Nizar Habash, Mohammad Salameh, Wajdi Zaghouani, Owen Rambow, Dana Abdulrahim, Ossama Obeid, Salam Khalifa, Fadhl Eryani, Alexander Erdmann, et al. The madar arabic dialect corpus and lexicon. In Proceedings of the eleventh international conference on language resources and evaluation (LREC 2018), 2018. find
  21. Mohamed Elmahdy, Mark Hasegawa-Johnson, and Eiman Mustafawi. Development of a tv broadcasts speech recognition system for qatari arabic. In LREC, volume 14, pages 3057–3061, 2014. find
  22. Hadry. French twitter sentiment analysis. https://kaggle.com/hbaflast/french-twitter-sentiment-analysis, 2020. Accessed: (August 15, 2023). ↗
  23. Ramy Baly, Alaa Khaddaj, Hazem Hajj, Wassim El-Hajj, and Khaled Bashir Shaban. Arsentd-lev: A multi-topic corpus for target-based sentiment analysis in arabic levantine tweets. arXiv preprint arXiv:1906.01830, 2019. ↗ our card
  24. Rana Abaalkhail, Fatimah Alzamzami, Samah Aloufi, Rajwa Alharthi, and Abdulmotaleb El Saddik. Affectional ontology and multimedia dataset for sentiment analysis. In International Conference on Smart Multimedia, pages 15–28. Springer, 2018. find
  25. Hatem Haddad, Hala Mulki, and Asma Oueslati. T-hsab: A tunisian hate speech and abusive dataset. In International Conference on Arabic Language Processing, pages 251–263. Springer, 2019. find
  26. Ibrahim Abu Farha and Walid Magdy. From arabic sentiment analysis to sarcasm detection: The arsarcasm dataset. In Proceedings of the 4th Workshop on Open-Source Arabic Corpora and Processing Tools, with a Shared Task on Offensive Language Detection, pages 32–39, 2020. find our card
  27. Muhammad Abdul-Mageed and Mona T Diab. Sana: A large scale multi-genre, multi-dialect lexicon for arabic subjectivity and sentiment analysis. In LREC, pages 1162–1169, 2014. find
  28. Hala Mulki and Bilal Ghanem. Let-mi: an arabic levantine twitter dataset for misogynistic language. arXiv preprint arXiv:2103.10195, 2021. ↗
  29. Maarten Grootendorst. Bertopic: Neural topic modeling with a class-based tf-idf procedure, 2022. find our card
  30. Sifatullah Siddiqi and Aditi Sharan. Keyword and keyphrase extraction techniques: a literature review. International Journal of Computer Applications, 109(2), 2015. find
  31. Rada Mihalcea and Paul Tarau. Textrank: Bringing order into text. In Proceedings of the 2004 conference on empirical methods in natural language processing, pages 404–411, 2004. find
  32. Xiaojun Wan and Jianguo Xiao. Single document keyphrase extraction using neighborhood knowledge. In AAAI, volume 8, pages 855–860, 2008. find
  33. Soheil Danesh, Tamara Sumner, and James H Martin. Sgrank: Combining statistical and graphical methods to improve the state of the art in unsupervised keyphrase extraction. In Proceedings of the fourth joint conference on lexical and computational semantics, pages 117–126, 2015. find
  34. Corina Florescu and Cornelia Caragea. Positionrank: An unsupervised approach to keyphrase extraction from scholarly documents. In Proceedings of the 55th annual meeting of the association for computational linguistics (volume 1: long papers), pages 1105–1115, 2017. find
  35. Amparo Elizabeth Cano Basave, Yulan He, and Ruifeng Xu. Automatic labelling of topic models learned from twitter by summarisation. In Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pages 618–624, 2014. find
  36. JIN Fenglei, Gao Cuiyun, et al. An online topic modeling framework with topics automatically labeled. In Proceedings of the 2019 Workshop on Widening NLP, pages 73–76, 2019. find
  37. Stuart Rose, Dave Engel, Nick Cramer, and Wendy Cowley. Automatic keyword extraction from individual documents. Text mining: applications and theory, 1:1–20, 2010. find
  38. Fatimah Alzamzami and Abdulmotaleb El Saddik. Osn-mdad: Machine translation dataset for arabic multi-dialectal conversations on online social media. arXiv preprint arXiv:2309.12137, 2023. ↗ our card
  39. Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li, Sergey Edunov, Marjan Ghazvininejad, Mike Lewis, and Luke Zettlemoyer. Multilingual denoising pre-training for neural machine translation. Transactions of the Association for Computational Linguistics, 8:726–742, 2020. find
  40. Alexander Rietzler, Sebastian Stabinger, Paul Opitz, and Stefan Engl. Adapt or get left behind: Domain adaptation through bert language model finetuning for aspect-target sentiment classification. arXiv preprint arXiv:1908.11860, 2019. ↗
  41. Fatimah Alzamzami and Abdulmotaleb El Saddik. Transformer-based feature fusion approach for multimodal visual sentiment recognition using tweets in the wild. IEEE Access, 2023. find
  42. Go Inoue, Bashar Alhafni, Nurpeiis Baimukan, Houda Bouamor, and Nizar Habash. The interplay of variant, size, and task type in Arabic pre-trained language models. In Proceedings of the Sixth Arabic Natural Language Processing Workshop, Kyiv, Ukraine (Online), April 2021. Association for Computational Linguistics. find our card
  43. Preslav Nakov, Sara Rosenthal, Zornitsa Kozareva, Veselin Stoyanov, Alan Ritter, and Theresa Wilson. SemEval-2013 task 2: Sentiment analysis in Twitter. In Second Joint Conference on Lexical and Computational Semantics (*SEM), Volume 2: Proceedings of the Seventh International Workshop on Semantic Evaluation (SemEval 2013), pages 312–320, Atlanta, Georgia, USA, June 2013. Association for Computational Linguistics. find our card
  44. Sara Rosenthal, Noura Farra, and Preslav Nakov. Semeval-2017 task 4: Sentiment analysis in twitter. In Proceedings of the 11th international workshop on semantic evaluation (SemEval-2017), pages 502–518, 2017. find our card
  45. Marwan Al Omari. Oclar: logistic regression optimisation for arabic customers’ reviews. International Journal of Business Intelligence and Data Mining, 20(3):251–273, 2022. find
  46. Dhuha Alqahtani, Lama Alzahrani, Maram Bahareth, Nora Alshameri, Hend Al-Khalifa, and Luluh Aldhubayi. Customer sentiments toward saudi banks during the covid-19 pandemic. In Proceedings of the 5th International Conference on Natural Language and Speech Processing (ICNLSP 2022), pages 251–257, 2022. find
  47. Sarah N Alyami and Sunday O Olatunji. Application of support vector machine for arabic sentiment classification using twitter-based dataset. Journal of Information & Knowledge Management, 19(01):2040018, 2020. find
  48. Valerio Basile, Cristina Bosco, Elisabetta Fersini, Nozza Debora, Viviana Patti, Francisco Manuel Rangel Pardo, Paolo Rosso, Manuela Sanguinetti, et al. Semeval-2019 task 5: Multilingual detection of hate speech against immigrants and women in twitter. In 13th International Workshop on Semantic Evaluation, pages 54–63. Association for Computational Linguistics, 2019. find our card
  49. Shorouq Zahra. Targeted topic modeling for levantine arabic, 2020. find
  50. Sahar Aldhaheri. Sentiment analysis for saudi public opinion toward covid19 and quarantine. https://www.linkedin.com/pulse/sentiment-analysis-saudi-public-opinion-toward-sahar-aldhaheri/, 2020. Accessed: (August 23, 2023). ↗
  51. Shaheen Syed and Marco Spruit. Full-text or abstract? examining topic coherence scores using latent dirichlet allocation. In 2017 IEEE International conference on data science and advanced analytics (DSAA), pages 165–174. IEEE, 2017. find
  52. Zuhe Li, Yangyu Fan, Bin Jiang, Tao Lei, and Weihua Liu. A survey on sentiment analysis and opinion mining for social multimedia. Multimedia Tools and Applications, 78:6939–6967, 2019. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/alzamzami-2023-content-localization-based-system-analyzing/