---
title: "Extending ImageNet to Arabic using Arabic WordNet (2020)"
description: "The paper asks how much of ImageNet (fall 2011 release, 21,841 synsets, about 14.2M images) can get Arabic labels from… 26 references."
authors: ["Abdulkareem Alsudais"]
year: 2020
venue: "Proceedings of the First Workshop on Advances in Language and Vision Research"
source: "https://aclanthology.org/2020.alvr-1.1/"
doi: "10.18653/v1/2020.alvr-1.1"
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/alsudais-2020-extending-imagenet-arabic-arabic-wordnet/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/alsudais-2020-extending-imagenet-arabic-arabic-wordnet/)

[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/10/) / Extending ImageNet to Arabic using Arabic WordNet

# Extending ImageNet to Arabic using Arabic WordNet

**Authors:** [Abdulkareem Alsudais](https://md.research.elamur.ai/free/en/authors/abdulkareem-alsudais/) · **Year:** 2020 · **Venue:** [Proceedings of the First Workshop on Advances in Language and Vision Research](https://md.research.elamur.ai/free/en/venues/alvr/) ([ALVR 2020](https://md.research.elamur.ai/free/en/venues/alvr/2020/)) · **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/2020.alvr-1.1/) · [PDF](https://aclanthology.org/2020.alvr-1.1.pdf) · [DOI 10.18653/v1/2020.alvr-1.1](https://doi.org/10.18653/v1/2020.alvr-1.1) · [OpenAlex](https://openalex.org/W3045491929)

**Citations:** 1 citations (OpenAlex, 2026-10-14)

## Summary
ImageNet labels its images with English WordNet synsets. No comparable large image collection with Arabic labels exists, which holds back Arabic vision-and-language work. The paper asks how much of ImageNet (fall 2011 release, 21,841 synsets, about 14.2M images) can get Arabic labels from Arabic WordNet (AWN), accessed through the Open Multilingual WordNet in NLTK and aligned on WordNet 3.0 IDs.

First, each ImageNet synset is looked up for a direct AWN equivalent. A recursive procedure then climbs every hypernym branch and records each Arabic synset it finds along with how many levels up it sits. Direct equivalents are scarce: only about 5.6% of synsets have one. Accepting ancestors raises coverage quickly, and with all levels allowed, only four synsets remain unlabeled, so 99.9% of images receive some Arabic label.

The author warns that labels taken from high levels become very generic. Gendered Arabic nouns, such as the AWN entry for "nurse" meaning a male nurse, and biased "person" categories also call for human evaluation.

## Key points
- Direct AWN matches cover 1,219 of 21,841 synsets, which accounts for 1,150,651 images (8.1%). The NLTK/OMW version of AWN holds only 9,916 Arabic synsets.
- Adding direct hypernyms raises coverage to 10,462 synsets and about 7.18M images (50.5%). Adding hypernyms of hypernyms reaches 17,438 synsets and 11,533,525 images (81.2%).
- With all hypernym levels, only four synsets lack an Arabic label, and they hold just 1,366 images. Some 7th-level ancestors are as broad as "entity", "act" and "event".
- Removing the 2,832-synset "person" subtree lowers coverage only slightly at every level.
- An earlier pilot found that machine-translated ImageNet labels were judged correct for only about 65% of a sample. A 100-image pilot with AWN found direct matches for just six synsets.

## Interesting ideas
- When no exact equivalent exists, walk up the hypernym graph to the closest ancestor that has one, and store the number of steps as a signal of how specific the label still is.
- Report coverage cumulatively per level and count it two ways, by distinct types (synsets) and by weighted mass (images). This shows how a fallback trades precision for reach.
- Rerun the whole pipeline with a sensitive or soon-to-be-removed subset excluded, to check that the headline numbers do not hinge on it.
- Fix the exact resource version, because synset IDs differ between WordNet 3.0 and 3.1, and a mismatch silently breaks alignment.
- Mapping labels across languages can inject grammatical gender that the source never stated, as with "nurse" becoming a male nurse.

## For Mova

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

- **I1** (world): Rare or unknown noun lemmas could inherit the class, measures and typical states of their nearest well-described WordNet ancestor, with the hop count kept as an explainable confidence value.
- **I2** (learn): Lexicon and ontology coverage reports could list exact-hit, one-hop and two-hop rates by both lemma type and corpus token frequency, which shows where back-off is doing too much of the work.
- **I3** (learn): Evaluation could routinely repeat scores with one domain or treebank slice held out, as a cheap control showing that gains are not carried by a single region of the data.
- **I4** (world): The compiled WordNet/FrameNet/Wikidata layer should record which WordNet release its offsets come from and refuse mixed-version joins.
- **I5** (back): Generation and translation through the meta-world should carry gender as an explicit, possibly unknown, entity attribute and not let a target-language default choose it.

*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

- [Prince Sattam Bin Abdulaziz University](https://md.research.elamur.ai/free/en/universities/prince-sattam-bin-abdulaziz-university/) — University, Al Kharj, Saudi Arabia · 17 works · 489 citations

📚 Shelves (5): [🖥️ 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 · [👁️ Computer Vision and Pattern Recognition](https://md.research.elamur.ai/free/en/shelves/computer-science/computer-vision-and-pattern-recognition/) · 2,869 works

🧭 Topics (1): [Semitic languages](https://md.research.elamur.ai/free/en/topics/semitic-languages/) · 2,908 works

## Cite

Cite as: Abdulkareem Alsudais. 2020. Extending ImageNet to Arabic using Arabic WordNet. Proceedings of the First Workshop on Advances in Language and Vision Research.

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

```bibtex
@inproceedings{alsudais2020extending,
  title = {{Extending ImageNet to Arabic using Arabic WordNet}},
  author = {Abdulkareem Alsudais},
  booktitle = {Proceedings of the First Workshop on Advances in Language and Vision Research},
  year = {2020},
  doi = {10.18653/v1/2020.alvr-1.1},
  url = {https://doi.org/10.18653/v1/2020.alvr-1.1}
}
```

- [← previous English WordNet Random Walk Pseudo-Corpora](https://md.research.elamur.ai/free/en/papers/klubicka-2020-english-wordnet-random-walk-pseudo/)
- [next → How to Start Teaching Amharic Verbs?](https://md.research.elamur.ai/free/en/papers/krzyzanowska-2020-start-teaching-amharic-verbs/)
- [↑ back to papers of 2020](https://md.research.elamur.ai/free/en/papers/2020/page/10/)

## References

26 references, 7 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Lahsen Abouenour, Karim Bouzoubaa, and Paolo Rosso. 2013. On the evaluation and improvement of Arabic WordNet coverage and usability. Language Resources and Evaluation, 47:891–917. [↗](https://doi.org/10.1007/s10579-013-9237-0)
2. Musa Alkhalifa and Horacio Rodríguez. 2009. Automatically extending NE coverage of Arabic WordNet using Wikipedia. In The 3rd International Conference on Arabic Language Processing CITALA2009., pages 20–36, Rabat, Morocco. [find](https://scholar.google.com/scholar?q=Musa+Alkhalifa+and+Horacio+Rodr%C3%ADguez%2E+2009%2E+Automatically+extending+NE+coverage+of+Arabic+WordNet+using+Wikipedia%2E+In+The+3rd+International+Conference+on+Arabic+Language+Processing+CITALA2009%2E%2C+pages+)
3. Huda A Al-muzaini, Tasniem N Al-yahya, and Hafida Benhidour. 2018. Automatic Arabic Image Captioning using RNN-LSTM-Based Language Model and CNN. International Journal of Advanced Computer Science and Applications, 9(6):67–73. [find](https://scholar.google.com/scholar?q=Huda+A+Al%2Dmuzaini%2C+Tasniem+N+Al%2Dyahya%2C+and+Hafida+Benhidour%2E+2018%2E+Automatic+Arabic+Image+Captioning+using+RNN%2DLSTM%2DBased+Language+Model+and+CNN%2E+International+Journal+of+Advanced+Computer+Science+and)
4. Abdulkareem Alsudais. 2019. [Image Classification in Arabic: Exploring Direct English to Arabic Translations](https://md.research.elamur.ai/free/en/papers/alsudais-2018-image-classification-arabic-assessing-accuracy/). IEEE Access, 7:122730–122739. [↗](https://doi.org/10.1109/access.2019.2926924)
5. Mohamed Ali Batita, Rami Ayadi, and Mounir Zrigui. 2019. Reasoning over Arabic WordNet Relations with Neural Tensor Network. Computación y Sistemas, 23(3):935–942. [↗](https://doi.org/10.13053/cys-23-3-3240)
6. William Black, Sabri Elkateb, Horacio Rodriguez, Musa Alkhalifa, Piek Vossen, Adam Pease, and Christiane Fellbaum. 2006. [Introducing the Arabic WordNet Project](https://md.research.elamur.ai/free/en/papers/bond-2013-linking-extending-open-multilingual-wordnet/). In Proceedings of the third international WordNet conference., pages 295–299.
7. Francis Bond and Kyonghee Paik. 2012. A Survey of WordNets and their Licenses. In Proceedings of the 6th Global WordNet Conference (GWC 2012), pages 64–71. [find](https://scholar.google.com/scholar?q=Francis+Bond+and+Kyonghee+Paik%2E+2012%2E+A+Survey+of+WordNets+and+their+Licenses%2E+In+Proceedings+of+the+6th+Global+WordNet+Conference+%28GWC+2012%29%2C+pages+64%E2%80%9371%2E)
8. Eva Cetinic, Tomislav Lipic, and Sonja Grgic. 2018. Fine-tuning Convolutional Neural Networks for fine art classification. Expert Systems With Applications, 114:107–118. [↗](https://doi.org/10.1016/j.eswa.2018.07.026)
9. Zhenfang Chen, Lin Ma, Wenhan Luo, and Kwan-Yee K.Wong. 2019. Weakly-Supervised Spatio-Temporally Grounding Natural Sentence in Video. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 1884–1894, Florence, Italy, July 28 - August 2, 2019. Association for Computational Linguistics. pages 1146–1157, Seattle,Washington, Association for Computational Linguistics. USA. [find](https://scholar.google.com/scholar?q=Zhenfang+Chen%2C+Lin+Ma%2C+Wenhan+Luo%2C+and+Kwan%2DYee+K%2EWong%2E+2019%2E+Weakly%2DSupervised+Spatio%2DTemporally+Grounding+Natural+Sentence+in+Video%2E+In+Proceedings+of+the+57th+Annual+Meeting+of+the+Association+for+)
10. Olga Russakovsky, Jia Deng, Hao Su, Jonathan Krause, Sanjeev Satheesh, Sean Ma, Zhiheng Huang, Andrej Karpathy, Aditya Khosla, Michael Bernstein, Alexander C Berg, and Li Fei-fei. 2015. [ImageNet large scale visual recognition challenge](https://md.research.elamur.ai/free/en/papers/russakovsky-2014-imagenet-large-scale-visual-recognition/). International Journal of Computer Vision, 115(3):211–252. [↗](https://doi.org/10.1007/s11263-015-0816-y)
11. Christopher Davis, Luana Bulat, Anita Vero, and Ekaterina Shutova. 2019. Deconstructing multimodality: visual properties and visual context in human semantic processing. In Proceedings of the Eighth Joint Conference on Lexical and Computational Semantics (*SEM), pages 118–124, Minneapolis, June 6–7, 2019. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Christopher+Davis%2C+Luana+Bulat%2C+Anita+Vero%2C+and+Ekaterina+Shutova%2E+2019%2E+Deconstructing+multimodality%3A+visual+properties+and+visual+context+in+human+semantic+processing%2E+In+Proceedings+of+the+Eighth+J)
12. Shreya Shankar, Yoni Halpern, Eric Breck, James Atwood, Jimbo Wilson, and D Sculley. 2017. No Classification without Representation: Assessing Geodiversity Issues in Open Data Sets for the Developing World. In NIPS 2017 workshop: Machine Learning for the Developing World. [↗](https://doi.org/10.48550/arxiv.1711.08536)
13. Jia Deng, Wei Dong, Richard Socher, Li-jia Li, Kai Li, and Li Fei-fei. 2009. ImageNet: A large-scale hierarchical image database. In 2009 IEEE Conference on Computer Vision and Pattern Recognition, pages 248–255, Miami, FL. [find](https://scholar.google.com/scholar?q=Jia+Deng%2C+Wei+Dong%2C+Richard+Socher%2C+Li%2Djia+Li%2C+Kai+Li%2C+and+Li+Fei%2Dfei%2E+2009%2E+ImageNet%3A+A+large%2Dscale+hierarchical+image+database%2E+In+2009+IEEE+Conference+on+Computer+Vision+and+Pattern+Recognition%2C+pa)
14. Ftoon Abu Shaqra, Rehab Duwairi, and Mahmoud Al-ayyoub. 2019. The Audio-Visual Arabic Dataset for Natural Emotions. In 2019 7th International Conference on Future Internet of Things and Cloud (FiCloud) The. IEEE. [find](https://scholar.google.com/scholar?q=Ftoon+Abu+Shaqra%2C+Rehab+Duwairi%2C+and+Mahmoud+Al%2Dayyoub%2E+2019%2E+The+Audio%2DVisual+Arabic+Dataset+for+Natural+Emotions%2E+In+2019+7th+International+Conference+on+Future+Internet+of+Things+and+Cloud+%28FiCloud)
15. Vasu Jindal. 2018. [Generating image captions in Arabic using root-word based recurrent neural networks and deep neural networks](https://md.research.elamur.ai/free/en/papers/jindal-2018-generating-image-captions-arabic-root/). In Proceedings of NAACL-HLT 2018: Student Research Workshop, pages 144–151.
16. Pierre Stock and Moustapha Cisse. 2018. ConvNets and ImageNet Beyond Accuracy: Understanding Mistakes and Uncovering Biases. In ECCV (6), pages 504–519. [find](https://scholar.google.com/scholar?q=Pierre+Stock+and+Moustapha+Cisse%2E+2018%2E+ConvNets+and+ImageNet+Beyond+Accuracy%3A+Understanding+Mistakes+and+Uncovering+Biases%2E+In+ECCV+%286%29%2C+pages+504%E2%80%93519%2E)
17. Alakananda Vempala and Daniel Preot. 2019. Categorizing and Inferring the Relationship between the Text and Image of Twitter Posts. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 2830–2840, Florence, Italy, July 28 - August 2, 2019. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Alakananda+Vempala+and+Daniel+Preot%2E+2019%2E+Categorizing+and+Inferring+the+Relationship+between+the+Text+and+Image+of+Twitter+Posts%2E+In+Proceedings+of+the+57th+Annual+Meeting+of+the+Association+for+Com)
18. Simon Kornblith, Jonathon Shlens, and Quoc V Le. 2019. Do Better ImageNet Models Transfer Better? In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 2661–2671. [find](https://scholar.google.com/scholar?q=Simon+Kornblith%2C+Jonathon+Shlens%2C+and+Quoc+V+Le%2E+2019%2E+Do+Better+ImageNet+Models+Transfer+Better%3F+In+Proceedings+of+the+IEEE+conference+on+computer+vision+and+pattern+recognition%2C+pages+2661%E2%80%932671%2E)
19. Ninareh Mehrabi, Fred Morstatter, Nripsuta Saxena, Kristina Lerman, and Aram Galstyan. 2019. [A Survey on Bias and Fairness in Machine Learning](https://md.research.elamur.ai/free/en/papers/mehrabi-2019-survey-bias-fairness-machine-learning/). arXiv:1908.09635v2. [↗](https://arxiv.org/abs/1908.09635)
20. Kaiyu Yang, Klint Qinami, Li Fei-Fei, Jia Deng, and Olga Russakovsky. 2020. Towards Fairer Datasets: Filtering and Balancing the Distribution of the People Subtree in the ImageNet Hierarchy. In Conference on Fairness, Accountability, and Transparency (FAT* ’20), pages 547–558, Barcelona, Spain. ACM. [find](https://scholar.google.com/scholar?q=Kaiyu+Yang%2C+Klint+Qinami%2C+Li+Fei%2DFei%2C+Jia+Deng%2C+and+Olga+Russakovsky%2E+2020%2E+Towards+Fairer+Datasets%3A+Filtering+and+Balancing+the+Distribution+of+the+People+Subtree+in+the+ImageNet+Hierarchy%2E+In+Confer)
21. George A Miller. 1995. [WordNet: A Lexical Database for English](https://md.research.elamur.ai/free/en/papers/miller-1994-wordnet/). Communications of the ACM, 38(11):39–41. [↗](https://doi.org/10.1145/219717.219748)
22. Finn Årup Nielsen. 2018. Linking ImageNet WordNet Synsets with Wikidata. In WWW ’18 Companion: The 2018 Web Conference Companion, pages 1809–1814, Lyon, France. [find](https://scholar.google.com/scholar?q=Finn+%C3%85rup+Nielsen%2E+2018%2E+Linking+ImageNet+WordNet+Synsets+with+Wikidata%2E+In+WWW+%E2%80%9918+Companion%3A+The+2018+Web+Conference+Companion%2C+pages+1809%E2%80%931814%2C+Lyon%2C+France%2E)
23. Mingyang Zhou, Runxiang Cheng, Yong Jae Lee, and Zhou Yu. 2018. A Visual Attention Grounding Neural Model for Multimodal Machine Translation. In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pages 3643–3653, Brussels, Belgium, October 31 - November 4, 2018. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Mingyang+Zhou%2C+Runxiang+Cheng%2C+Yong+Jae+Lee%2C+and+Zhou+Yu%2E+2018%2E+A+Visual+Attention+Grounding+Neural+Model+for+Multimodal+Machine+Translation%2E+In+Proceedings+of+the+2018+Conference+on+Empirical+Methods)
24. Princeton University. 2010. About WordNet. Princeton University, Princeton, NJ 08544, USA. [find](https://scholar.google.com/scholar?q=Princeton+University%2E+2010%2E+About+WordNet%2E+Princeton+University%2C+Princeton%2C+NJ+08544%2C+USA%2E)
25. Yasser Regragui, Lahsen Abouenour, Fettoum Krieche, Karim Bouzoubaa, and Paolo Rosso. 2016. [Arabic WordNet: New content and new applications](https://md.research.elamur.ai/free/en/papers/regragui-2016-arabic-wordnet-new-content-new/) Arabic WordNet. In Proceedings of the Eighth Global WordNet Conference, number January, pages 330–338.
26. Stephen Roller and Sabine Schulte. 2013. A Multimodal LDA Model Integrating Textual, Cognitive and Visual Modalities. In Proceedings of the 2013 Conference on Empirical Methods in Natural Language Processing, number October, [find](https://scholar.google.com/scholar?q=Stephen+Roller+and+Sabine+Schulte%2E+2013%2E+A+Multimodal+LDA+Model+Integrating+Textual%2C+Cognitive+and+Visual+Modalities%2E+In+Proceedings+of+the+2013+Conference+on+Empirical+Methods+in+Natural+Language+Pro)

*Updated 1 November 2026*