---
title: "NileULex — paper by El-Beltagy (2016) — summary and ideas"
description: "El-Beltagy presents NileULex, a manually curated lexicon of 5953 single words and compound phrases. 21 references, cited by 13 papers here."
authors: ["Samhaa R. El-Beltagy"]
year: 2016
venue: "Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC'16)"
source: "https://aclanthology.org/L16-1463/"
licence: "CC BY-NC 4.0 (ELRA, LREC proceedings)"
licence_url: "https://creativecommons.org/licenses/by-nc/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/beltagy-2016-nileulex-phrase-word-level-sentiment/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/beltagy-2016-nileulex-phrase-word-level-sentiment/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2016](https://md.research.elamur.ai/free/en/papers/2016/page/3/) / NileULex: A Phrase and Word Level Sentiment Lexicon for Egyptian and Modern Standard Arabic

# NileULex: A Phrase and Word Level Sentiment Lexicon for Egyptian and Modern Standard Arabic

**Authors:** [Samhaa R. El-Beltagy](https://md.research.elamur.ai/free/en/authors/samhaa-r-el-beltagy/) · **Year:** 2016 · **Venue:** [Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC'16)](https://md.research.elamur.ai/free/en/venues/lrec/) ([LREC 2016](https://md.research.elamur.ai/free/en/venues/lrec/2016/)) · **Publisher:** [European Language Resources Association (ELRA)](https://md.research.elamur.ai/free/en/publishers/european-language-resources-association-elra/) · **Licence:** [CC BY-NC 4.0 (ELRA, LREC proceedings)](https://creativecommons.org/licenses/by-nc/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/L16-1463/) · [PDF](https://aclanthology.org/L16-1463.pdf) · [OpenAlex](https://openalex.org/W2575929989)

**Citations:** 52 citations (OpenAlex, 2026-10-14) · [13](https://md.research.elamur.ai/free/en/papers/beltagy-2016-nileulex-phrase-word-level-sentiment/cited-by/) in our library, of the 149,439 papers whose bibliographies we read

## Summary
Arabic social media is mostly written in colloquial dialects, where sentiment is often carried by multiword idioms. Yet in 2016 almost no public Arabic sentiment lexicons existed. The few available focused on Modern Standard Arabic (MSA), ignored phrases, or, like EmoLex, were machine-translated from English. El-Beltagy presents NileULex, a manually curated lexicon of 5953 single words and compound phrases. About 45% of the entries are Egyptian or colloquial and 55% are MSA.

Candidate entries were often harvested automatically, but every entry was checked by a native speaker. Ambiguous items were removed or replaced by the longer phrases in which their polarity is clear.

The lexicon was tested by adding a few lexicon-count features to a unigram/bigram TF-IDF Complement Naïve Bayes classifier. Experiments used Egyptian and Saudi Twitter data. NileULex raised accuracy over the no-lexicon baseline and outperformed the translated EmoLex. The work shows that a small, hand-validated, phrase-aware resource can beat larger automatic or translated ones for dialectal text.

## Key points
- **Lexicon composition:** NileULex has 5953 unique entries:
  - 3693 single negative words and 1281 single positive words;
  - 563 negative and 416 positive compound phrases.
- **Dialect split:** 2674 entries are Egyptian/colloquial and 3279 are MSA. The lexicon also includes:
  - a few terms from other dialects;
  - transliterated English words, such as words for "cute" and "like";
  - common misspellings of frequent words.
- **Origins:** 67.6% of entries come from the earlier lexicon of El-Beltagy & Ali (2013). NileULex overlaps with EmoLex v0.92 (4187 unique entries) by 973 terms.
- **Classifier setup:** Complement Naïve Bayes in Weka, with lemmatization done by a modified El-Beltagy & Rafea stemmer. The added features were:
  - counts of positive and negative lexicon hits;
  - the polarity of the last matched entry;
  - a weight of 1.5 for compound-phrase hits instead of 1.
- **Results on the 2-class Egyptian dataset (NU_EG_Twitter_corpus):**

  | Setting | Baseline | NileULex | EmoLex | Combined |
  |---|---|---|---|---|
  | 10-fold CV accuracy | 82.59% | 85.4% | 84.22% | — |
  | Held-out test accuracy | 73.94% | 77.58% | 76.77% | 77.98% |

## Interesting ideas
- When a frequent word (such as "Allah" used like "wow") can go either way, drop it and list the full phrases it starts, each with its own polarity; this avoids guessing from an unreliable head word.
- Most listed idioms have no polarity signal in their parts, so the lexicon treats them as whole units and counts phrase matches more heavily than word matches.
- For lexicon lookup, lemmatize instead of stemming, because stripping a single character can flip a word's meaning (in the paper's example, "magnificent" becomes "terrorized").
- The study keeps the classifier deliberately simple and leaves out emoticons and other cues, so any gain can be credited to the lexicon alone. It also compares against a machine-translated lexicon as a control.
- The paper notes that negation does not always flip polarity: the Egyptian phrase meaning "no, nice" affirms niceness.

## For Mova

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

- **I1** (world): Compiled knowledge entries for polarity or speech-act cues could be keyed on whole constructions rather than on ambiguous single lemmas, so each lookup stays explainable and unambiguous.
- **I2** (frames): Irony and implicature detection in prag could include a multiword-expression lexicon matched against UD subtrees, with phrase-level evidence outranking word-level evidence.
- **I3** (tokens): The en crate should match knowledge resources on lemmas, never on aggressive stems, and could add a test that flags stem collisions which change meaning.
- **I4** (learn): When evaluating a new resource, ablate everything else and include a "cheap substitute" control (e.g. an auto-translated or auto-induced version) to show the resource itself adds value.
- **I5** (frames): Polarity-flip rules should be overridable by listed constructions, so negation handling in frames treats such idioms as exceptions rather than applying a blanket rule.

*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

- [Nile University](https://md.research.elamur.ai/free/en/universities/nile-university/) — University, Giza, Egypt · 14 works · 158 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: Samhaa R. El-Beltagy. 2016. NileULex: A Phrase and Word Level Sentiment Lexicon for Egyptian and Modern Standard Arabic. Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC'16).

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

```bibtex
@inproceedings{elbeltagy2016nileulex,
  title = {{NileULex: A Phrase and Word Level Sentiment Lexicon for Egyptian and Modern Standard Arabic}},
  author = {Samhaa R. El-Beltagy},
  booktitle = {Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC'16)},
  year = {2016},
  doi = {10.63317/44ooe3eubfv6},
  url = {https://doi.org/10.63317/44ooe3eubfv6}
}
```

- [← previous Idiom Token Classification using Sentential Distributed Semantics](https://md.research.elamur.ai/free/en/papers/salton-2016-idiom-token-classification-sentential-distributed/)
- [next → A Large Scale Corpus of Gulf Arabic](https://md.research.elamur.ai/free/en/papers/khalifa-2016-large-scale-corpus-gulf-arabic/)
- [↑ back to papers of 2016](https://md.research.elamur.ai/free/en/papers/2016/page/3/)

## References

21 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, Muhammad, and Mona Diab. 2012. “Toward Building a Large-Scale Arabic Sentiment Lexicon.” In Proceedings of the 6th International Global WordNet Conference, 18–22. Matuse, Japan. [find](https://scholar.google.com/scholar?q=Abdul%2DMageed%2C+Muhammad%2C+and+Mona+Diab%2E+2012%2E+%E2%80%9CToward+Building+a+Large%2DScale+Arabic+Sentiment+Lexicon%2E%E2%80%9D+In+Proceedings+of+the+6th+International+Global+WordNet+Conference%2C+18%E2%80%9322%2E+Matuse%2C+Japan%2E)
2. Al-Sabbagh, R, and Roxana Girju. 2010. “[Mining the Web for the Induction of a Dialectical Arabic Lexicon](https://md.research.elamur.ai/free/en/papers/sabbagh-2010-mining-web-induction-dialectical-arabic/).” In LREC. European Language Resources …, 288–93. [http://www.lrec-conf.org/proceedings/lrec2010/pdf/344_Paper.pdf](http://www.lrec-conf.org/proceedings/lrec2010/pdf/344_Paper.pdf). [↗](http://www.lrec-conf.org/proceedings/lrec2010/pdf/344_Paper.pdf)
3. Baccianella, Stefano, 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 Proceedings of the Seventh International Conference on Language Resources and Evaluation (LREC’10), 0:2200–2204. doi:citeulike-article-id:9238846.
4. Badaro, G., R. Baly, H. Hajj, N. Habash, and W. El-Hajj. 2014. “[A Large Scale Arabic Sentiment Lexicon for Arabic Opinion Mining](https://md.research.elamur.ai/free/en/papers/badaro-2014-large-scale-arabic-sentiment-lexicon/).” In Proceedings of the EMNLP Workshop on Arabic Natural Language Processing (ANLP), 165–73. Association for Computational Linguistics.
5. El-Beltagy, Samhaa R, and Ahmed Ali. 2013. “Open Issues in the Sentiment Analysis of Arabic Social Media : A Case Study.” In Proceedings of 9th the International Conference on Innovations and Information Technology (IIT2013). Al Ain, UAE. [find](https://scholar.google.com/scholar?q=El%2DBeltagy%2C+Samhaa+R%2C+and+Ahmed+Ali%2E+2013%2E+%E2%80%9COpen+Issues+in+the+Sentiment+Analysis+of+Arabic+Social+Media+%3A+A+Case+Study%2E%E2%80%9D+In+Proceedings+of+9th+the+International+Conference+on+Innovations+and+Informat)
6. El-Beltagy, Samhaa R., and Ahmed Rafea. 2011. “An Accuracy Enhanced Light Stemmer for Arabic Text.” ACM Transactions on Speech and Language Processing 7 (2): 2–23. [↗](https://doi.org/10.1145/1921656.1921657)
7. El-Beltagy, Samhaa R., and Ahmed Rafea. 2016. “LemaLight: A Dictionary Based Arabic Lemmatizer and Stemmer.” TR2-11-16, Technical report.. [find](https://scholar.google.com/scholar?q=El%2DBeltagy%2C+Samhaa+R%2E%2C+and+Ahmed+Rafea%2E+2016%2E+%E2%80%9CLemaLight%3A+A+Dictionary+Based+Arabic+Lemmatizer+and+Stemmer%2E%E2%80%9D+TR2%2D11%2D16%2C+Technical+report%2E%2E)
8. El-Bletagy, Samhaa R. “NileTMRG: Deriving Prior Polarities for Arabic Sentiment Terms.” In Proceedings of SemEval 2016 -(to Appear). San Diego, California. [find](https://scholar.google.com/scholar?q=El%2DBletagy%2C+Samhaa+R%2E+%E2%80%9CNileTMRG%3A+Deriving+Prior+Polarities+for+Arabic+Sentiment+Terms%2E%E2%80%9D+In+Proceedings+of+SemEval+2016+%2D%28to+Appear%29%2E+San+Diego%2C+California%2E)
9. ElSahar, Hady, and Samhaa R. El-Beltagy. 2014. “A Fully Automated Approach for Arabic Slang Lexicon Extraction from Microblogs”.” Lecture Notes in Computer Science (LNCS) - CICLing 2014, (Editor A. Glbukh) 8403. [find](https://scholar.google.com/scholar?q=ElSahar%2C+Hady%2C+and+Samhaa+R%2E+El%2DBeltagy%2E+2014%2E+%E2%80%9CA+Fully+Automated+Approach+for+Arabic+Slang+Lexicon+Extraction+from+Microblogs%E2%80%9D%2E%E2%80%9D+Lecture+Notes+in+Computer+Science+%28LNCS%29+%2D+CICLing+2014%2C+%28Editor+A%2E+Gl)
10. “Facebook Statistics by Country.” 2012. [http://www.socialbakers.com/facebook-statistics/](http://www.socialbakers.com/facebook-statistics/). [↗](http://www.socialbakers.com/facebook-statistics/)
11. Farid, Doaa. 2013. “Egypt Has the Largest Number of Facebook Users in the Arab World.” Daily News Egypt, September 23. [http://www.dailynewsegypt.com/2013/09/25/egypt-has-the-largest-number-of-facebook-users-in-the-ara](http://www.dailynewsegypt.com/2013/09/25/egypt-has-the-largest-number-of-facebook-users-in-the-ara) b-world-report/. [↗](http://www.dailynewsegypt.com/2013/09/25/egypt-has-the-largest-number-of-facebook-users-in-the-ara)
12. Frank, Eibe, Mark Hall, Geoffrey Holmes, Richard Kirkby, Bernhard Pfahringer, Ian H Witten, and Len Trigg. 2005. “WEKA: A Machine Learning Workbench for Data Mining.” In Data Mining and Knowledge Discovery Handbook, edited by Oded Maimon and Lior Rokach, 1305–14. Springer. doi:10.1007/0-387-25465-X_62. [↗](https://doi.org/10.1007/0-387-25465-x_62)
13. Khalil, Talaat, Amal Halaby, Muhammad H. Hammad, and Samhaa R. El-Beltagy. 2015. “Which Configuration Works Best? An Experimental Study on Supervised Arabic Twitter Sentiment Analysis.” In Proceedings of the First Conference on Arabic Computational Liguistics (ACLing 2015), Co-Located with CICLing 2015 (to Appear). Cairo, Egypt. [find](https://scholar.google.com/scholar?q=Khalil%2C+Talaat%2C+Amal+Halaby%2C+Muhammad+H%2E+Hammad%2C+and+Samhaa+R%2E+El%2DBeltagy%2E+2015%2E+%E2%80%9CWhich+Configuration+Works+Best%3F+An+Experimental+Study+on+Supervised+Arabic+Twitter+Sentiment+Analysis%2E%E2%80%9D+In+Proceedings)
14. Kiritchenko, Svetlana, Xiaodan Zhu, and Saif Mohammad. 2014. “Sentiment Analysis of Short Informal Texts.” Journal of Artificial Intelligence Research 50: 723– 62. [↗](https://doi.org/10.1613/jair.4272)
15. Liu, Bing. 2010. “Sentiment Analysis and Subjectivity.” In Handbook of Natural Language Processing, Second Edition, edited by N. Indurkhya and F. J. Damerau. [find](https://scholar.google.com/scholar?q=Liu%2C+Bing%2E+2010%2E+%E2%80%9CSentiment+Analysis+and+Subjectivity%2E%E2%80%9D+In+Handbook+of+Natural+Language+Processing%2C+Second+Edition%2C+edited+by+N%2E+Indurkhya+and+F%2E+J%2E+Damerau%2E)
16. Mahyouba, Fawaz H.H., Muazzam A. Siddiquia, and Mohamed Y. Dahaba. 2014. “Building an Arabic Sentiment Lexicon Using Semi-Supervised Learning.” Journal of King Saud University - Computer and Information Sciences 26 (4): 417–24. [find](https://scholar.google.com/scholar?q=Mahyouba%2C+Fawaz+H%2EH%2E%2C+Muazzam+A%2E+Siddiquia%2C+and+Mohamed+Y%2E+Dahaba%2E+2014%2E+%E2%80%9CBuilding+an+Arabic+Sentiment+Lexicon+Using+Semi%2DSupervised+Learning%2E%E2%80%9D+Journal+of+King+Saud+University+%2D+Computer+and+Informati)
17. Mohammad, Saif, and Peter Turney. 2013. “Crowdsourcing a Word-Emotion Association Lexicon.” Computational Intelligence 29 (3): 436– 65. [↗](https://doi.org/10.1111/j.1467-8640.2012.00460.x)
18. Neal, Ryan W. 2013. “Twitter Usage Statistics: Which Country Has The Most Active Twitter Population?” International Business Times. [http://www.ibtimes.com/twitter-usage-statistics-whi](http://www.ibtimes.com/twitter-usage-statistics-whi) ch-country-has-most-active-twitter-population-1474 852. [↗](http://www.ibtimes.com/twitter-usage-statistics-whi)
19. Rennie, Jason D M, Lawrence Shih, Jaime Teevan, and David R Karger. 2003. “Tackling the Poor Assumptions of Naive Bayes Text Classifiers.” Proceedings of the Twentieth International Conference on Machine Learning (ICML)-2003) 20 (1973): 616–23. doi:10.1186/1477-3155-8-16. [↗](https://doi.org/10.1186/1477-3155-8-16)
20. Salton, Gerard, and Chris Buckley. 1988. “Term-Weighting Approaches in Automatic Text Retrieval.” Information Processing and Management 24 (5): 513–23. [↗](https://doi.org/10.1016/0306-4573(88)90021-0)
21. Wilson, Theresa, Janyce Wiebe, and Paul Hoffmann. 2005. “[Recognizing Contextual Polarity in Phrase-Level Sentiment Analysis](https://md.research.elamur.ai/free/en/papers/wilson-2005-recognizing-contextual-polarity-phrase-level/).” In Proceedings of Human Language Technology Conference and Conference on Empirical Methods in Natural Language Processing (HLT/EMNLP), 347–54. Vancouver, Canada.

*Updated 1 November 2026*