--- viewer: false language: - ara tags: - Arabic - NLU Benchmark - Natural Language Inference (NLI) - Question Answering (QA) - Semantic Textual Similarity and and Paraphrase (STSP) - Sentence Classification (SC) - Structure Predictions (SP) - Topic Classification (TC) - Word Sense Disambiguation (WSD) task_categories: - text-classification - token-classification - question-answering extra_gated_fields: Name: text Official Email (email of your organization): text Affilation: text Country: text I agree to use this dataset for non-commercial use ONLY: checkbox I agree to cite the ORCA paper and all original papers: checkbox ---



Documentation

In this work, we introduce [**ORCA**](https://arxiv.org/abs/2212.10758), a publicly available benchmark for Arabic language understanding evaluation. ORCA is carefully constructed to cover diverse Arabic varieties and a wide range of challenging Arabic understanding tasks exploiting 60 different datasets across seven NLU task clusters. To measure current progress in Arabic NLU, we use ORCA to offer a comprehensive comparison between 18 multilingual and Arabic language models. # ORCA Task Cluster We arrange [**ORCA**](https://arxiv.org/abs/2212.10758), into seven NLU task clusters. These are (1) sentence classification, (2) structured prediction (3) semantic textual similarity and paraphrase, (4) text classification, (5) natural language inference, (6) word sense disambiguation, and (7) question answering. ### (1) Natural Language Inference (NLI) |**Task**| **Variation** | **Metric** | **Reference** | |---------|--------|--------|------| |[ANS Stance](https://aclanthology.org/2020.fever-1.2/) |MSA | Macro F1 | [(Khouja, 2020)](https://aclanthology.org/2020.fever-1.2/) | |[Baly Stance](https://aclanthology.org/N18-2004/) |MSA | Macro F1 | [(Balyet al., 2018)](https://aclanthology.org/N18-2004/) | |[XLNI](https://github.com/facebookresearch/XNLI) |MSA | Macro F1 | [(Conneau et al., 2018)](https://github.com/facebookresearch/XNLI)| ### (2) Question Answering (QA) |**Task**| **Variation** | **Metric** | **Reference** | |---------|--------|--------|------| |[Question Answering](https://aclanthology.org/2021.acl-long.551/) |MSA | Macro F1 | [(Abdul-Mageed et al., 2020a)](https://aclanthology.org/2021.acl-long.551/) | ### (3) Semantic Textual Similarity and Paraphrase (STSP) |**Task**| **Variation** | **Metric** | **Reference** | |---------|--------|--------|-------| |[Emotion Regression](https://aclanthology.org/S18-1001/) |MSA | Spearman Correlation| [(Saif et al., 2018)](https://aclanthology.org/S18-1001/) | |[MQ2Q](https://aclanthology.org/2019.nsurl-1.1) |MSA | Macro F1 | [(Seelawi al., 2019)](https://aclanthology.org/2019.nsurl-1.1) | |[STS](https://aclanthology.org/S17-2001/) |MSA | Macro F1 | [(Cer et al., 2017)](https://aclanthology.org/S17-2001/) | ### (4) Sentence Classification (SC) |**Task**| **Variation** | **Metric** | **Reference** | |---------|--------|--------|-------| |[Abusive](https://aclanthology.org/W19-3512/) |DA | Macro F1 | [(Mulki et al., 2019)](https://aclanthology.org/W19-3512/) | |[Adult](https://aclanthology.org/2021.wanlp-1.14) |DA | Macro F1 | [(Mubarak et al., 2021)](https://aclanthology.org/2021.wanlp-1.14) | |[Age](https://www.aclweb.org/anthology/2020.osact-1.3) |DA | Macro F1 | [(Abdul-Mageed et al., 2020b)]( https://aclanthology.org/2020.osact-1.3/) | |[ANS Claim](https://aclanthology.org/2020.fever-1.2/) |MSA | Macro F1 | [(Khouja, 2020)](https://aclanthology.org/2020.fever-1.2/) | |[Dangerous ](https://aclanthology.org/N18-2004/) |DA | Macro F1 | [(Alshehri et al., 2020)](https://www.aclweb.org/anthology/2020.osact-1.6)| |[Dialect Binary](https://github.com/facebookresearch/XNLI) |DA | Macro F1 | [(Farha, 2020)](https://aclanthology.org/2020.osact-1.5/), [(Zaidan, 2014)](https://www.aclweb.org/anthology/J14-1006), [(Abdul-Mageed et al., 2020c)](https://aclanthology.org/2021.acl-long.551/), [(Bouamor et al., 2019)](https://www.aclweb.org/anthology/W19-4622), [(Abdelaliet al., 2020)](https://aclanthology.org/2021.wanlp-1.1), [(El-Haj, 2020)](https://aclanthology.org/2020.lrec-1.165/). | |[Dialect Country](https://github.com/facebookresearch/XNLI) |DA | Macro F1 | [(Farha, 2020)](https://aclanthology.org/2020.osact-1.5/), [(Zaidan, 2014)](https://www.aclweb.org/anthology/J14-1006), [(Abdul-Mageed et al., 2020c)](https://aclanthology.org/2021.acl-long.551/), [(Bouamor et al., 2019)](https://www.aclweb.org/anthology/W19-4622), [(Abdelaliet al., 2020)](https://aclanthology.org/2021.wanlp-1.1), [(El-Haj, 2020)](https://aclanthology.org/2020.lrec-1.165/). | |[Dialect Region](https://github.com/facebookresearch/XNLI) |DA | Macro F1 | [(Farha, 2020)](https://aclanthology.org/2020.osact-1.5/), [(Zaidan, 2014)](https://www.aclweb.org/anthology/J14-1006), [(Abdul-Mageed et al., 2020c)](https://aclanthology.org/2021.acl-long.551/), [(Bouamor et al., 2019)](https://www.aclweb.org/anthology/W19-4622), [(Abdelaliet al., 2020)](https://aclanthology.org/2021.wanlp-1.1), [(El-Haj, 2020)](https://aclanthology.org/2020.lrec-1.165/). | |[Emotion](https://www.aclweb.org/anthology/2020.osact-1.3) |DA | Macro F1 | [(Abdul-Mageed et al., 2020b)]( https://aclanthology.org/2020.osact-1.3/) | |[Gender](https://www.aclweb.org/anthology/2020.osact-1.3) |DA | Macro F1 | [(Abdul-Mageed et al., 2020b)]( https://aclanthology.org/2020.osact-1.3/) | |[Hate Speech](https://www.aclweb.org/anthology/2020.osact-1.7) |DA | Macro F1 | [(Mubarak et al., 2020)](https://www.aclweb.org/anthology/2020.osact-1.7)| |[Irony](https://dl.acm.org/doi/10.1145/3368567.3368585) |DA | Macro F1 | [(Ghanem al., 2019)](https://dl.acm.org/doi/10.1145/3368567.3368585) | |[Machine Generation](https://aclanthology.org/2020.wanlp-1.7/) |MSA | Macro F1 | [(Nagoudi et al., 2020)](https://aclanthology.org/2020.wanlp-1.7/) | |[Offensive](https://aclanthology.org/2020.osact-1.8/) |DA | Macro F1 | [(Mubarak et al., 2020)](https://www.aclweb.org/anthology/2020.osact-1.7)| |[Sarcasm](https://aclanthology.org/N18-2004/) |DA | Macro F1 | [(Farha and Magdy, 2020)](https://aclanthology.org/2020.osact-1.5/) | |[Sentiment Analysis](https://aclanthology.org/2021.acl-long.551/) |DA | Macro F1 | [(Abdul-Mageed et al., 2020c)](https://aclanthology.org/2021.acl-long.551/) | ### (5) Structure Predictions (SP) |**Task**| **Variation** | **Metric** | **Reference** | |---------|--------|--------|-------| |[Aqmar NER](https://www.cs.cmu.edu/~ark/ArabicNER/) |MSA | Macro F1 | [(Mohit, 2012)](https://www.cs.cmu.edu/~ark/ArabicNER/) | |[Arabic NER Corpus](http://www.dsic.upv.es/~prosso/resources/BenajibaRosso_IICAI07.pdf) |MSA | Macro F1 | [(Benajiba and Rosso, 2007)](http://www.dsic.upv.es/~prosso/resources/BenajibaRosso_IICAI07.pdf) | |[Dialect Part Of Speech](https://aclanthology.org/L18-1015.pdf) |DA | Macro F1 | [(Darwish et al., 2018)](https://aclanthology.org/L18-1015.pdf) | |[MSA Part Of Speech](https://arxiv.org/abs/2004.01401) |MSA | Macro F1 | [(Liang et al., 2020)](https://arxiv.org/abs/2004.01401) | ### (6) Topic Classification (TC) |**Task**| **Variation** | **Metric** | **Reference** | |---------|--------|--------|-------| |[Topic](https://aclanthology.org/2021.acl-long.551/) |MSA | Macro F1 | [(Abbas et al.,2011)](https://www.dline.info/fpaper/jdim/v9i5/1.pdf), [(Chouigui et al.,2017)](https://www.researchgate.net/publication/320871871_Poster_ANT_Corpus_An_Arabic_News_Text_Collection_for_Textual_Classification), [(Saad, 2010)](http://site.iugaza.edu.ps/wp-content/uploads/mksaad-OSAC-OpenSourceArabicCorpora-EECS10-rev9(1).pdf). | ### (7) Word Sense Disambiguation (WSD) |**Task**| **Variation** | **Metric** | **Reference** | |---------|--------|--------|-------| |[Word Sense Disambiguation](https://www.mdpi.com/2076-3417/11/6/2567) |MSA | Macro F1 | [(El-Razzaz, 2021)](https://www.mdpi.com/2076-3417/11/6/2567) | # How to Use ORCA ### Request Access ### To obtain access to the ORCA benchmark on Huggingface, follow the following steps: - Login on your Haggingface account - Request access ### Install Requirments ```shell pip install datasets transformers seqeval ``` ### Login with your Huggingface CLI ### You can get/manage your access tokens in your [settings](https://huggingface.co/docs/hub/security-tokens). ```shell export HUGGINGFACE_TOKEN="" huggingface-cli login --token $HUGGINGFACE_TOKEN ``` ### Fine-tuning a model on ORCA tasks We provide a Google Colab Notebook that includes instructions for fine-tuning any model on ORCA tasks. colab ### Submitting your results on ORCA test We design a public leaderboard for scoring PLMs on ORCA. Our leaderboard is interactive and offers rich meta-data about the various datasets involved as well as the language models we evaluate. You can evalute your models using **ORCA** leaderboard: **[https://orca.dlnlp.ai](https://orca.dlnlp.ai/)** --- ## Citation If you use ORCA for your scientific publication, or if you find the resources in this repository useful, please cite our paper as follows: ``` @inproceedings{elmadany-etal-2023-orca, title = "{ORCA}: A Challenging Benchmark for {A}rabic Language Understanding", author = "Elmadany, AbdelRahim and Nagoudi, ElMoatez Billah and Abdul-Mageed, Muhammad", booktitle = "Findings of the Association for Computational Linguistics: ACL 2023", month = jul, year = "2023", address = "Toronto, Canada", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2023.findings-acl.609", pages = "9559--9586", } ``` --- ## Acknowledgments We gratefully acknowledge support from the Natural Sciences and Engineering Research Council of Canada, the Social Sciences and Humanities Research Council of Canada, Canadian Foundation for Innovation, [ComputeCanada](www.computecanada.ca) and [UBC ARC-Sockeye](https://doi.org/10.14288/SOCKEYE). We also thank the [Google TensorFlow Research Cloud (TFRC)](https://www.tensorflow.org/tfrc) program for providing us with free TPU access.