Name
stringlengths 3
106
| Subsets
list | HF Link
stringlengths 3
69
| Link
stringlengths 24
135
| License
stringclasses 28
values | Year
int32 2k
2.02k
| Language
stringclasses 2
values | Dialect
stringclasses 18
values | Domain
stringclasses 10
values | Form
stringclasses 3
values | Collection Style
stringclasses 7
values | Description
stringlengths 16
1.64k
| Volume
stringlengths 1
13
| Unit
stringclasses 5
values | Ethical Risks
stringclasses 4
values | Provider
stringlengths 1
136
| Derived From
stringlengths 2
307
| Paper Title
stringlengths 3
143
| Paper Link
stringlengths 3
285
| Script
stringclasses 4
values | Tokenized
stringclasses 2
values | Host
stringclasses 21
values | Access
stringclasses 3
values | Cost
stringlengths 3
11
| Test Split
stringclasses 3
values | Tasks
stringlengths 8
181
| Venue Title
stringlengths 2
46
| Citations
stringlengths 3
6
| Venue Type
stringclasses 5
values | Venue Name
stringlengths 3
113
| Authors
stringlengths 3
923
| Affiliations
stringlengths 1
470
| Abstract
stringlengths 3
2.15k
| Added By
stringlengths 3
25
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Arabic-Poem-Emotion | [] | Arabic-Poem-Emotion | https://github.com/SakibShahriar95/Arabic-Poem-Emotion | unknown | 2,021 | ar | mixed | other | text | crawling | A dataset containing over 9000 Arabic poems labeled by three emotion classes. | 9,000 | sentences | Low | University of Guelph | nan | nan | nan | Arab | No | GitHub | Free | nan | No | poem emotion Classification | nan | nan | nan | nan | nan | nan | nan | Abdelrahman Rezk |
Sudannese Arabic Telcom Sentiment Classification Pre Processed | [] | nan | https://docs.google.com/spreadsheets/d/13fIV8oHss-QRBKN-2h5LYF1i_1O9qH1R/edit?usp=sharing&ouid=113975694262803649646&rtpof=true&sd=true | MIT License | 2,018 | ar | ar-SD: (Arabic (Sudan)) | other | text | crawling and annotation(other) | it is pre processed dataset from Twitter about Telecom companies in Sudan, it labelled by 3 different labels from different age, gender and background | 5,349 | sentences | High | University of Khartoum | nan | Sentiment analysis for arabic dialect using supervised learning | https://ieeexplore.ieee.org/document/8515862 | Latn | No | Gdrive | Upon-Request | nan | No | sentiment analysis | nan | 14.0 | conference | nan | nan | nan | nan | Rua Ismail |
OrienTel Egypt MSA (Modern Standard Arabic) database | [] | nan | https://catalogue.elra.info/en-us/repository/browse/ELRA-S0222/ | Non Commercial Use - ELRA END USER | 2,006 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | transcribed audio | spoken | other | The OrienTel Egypt MSA (Modern Standard Arabic) database comprises 500 Egyptian speakers (254 males, 246 females) recorded over the Egyptian fixed and mobile telephone network. | 24,500 | sentences | Low | nan | nan | nan | Arab | No | ELRA | With-Fee | 15,000.00€ | No | speech recognition | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
|
OrienTel Jordan MSA (Modern Standard Arabic) database | [] | nan | https://catalogue.elra.info/en-us/repository/browse/ELRA-S0290/ | Non Commercial Use - ELRA END USER | 2,008 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | transcribed audio | spoken | other | The OrienTel Jordan MSA (Modern Standard Arabic) database comprises 556 Jordanian speakers (288 males, 268 females) recorded over the Jordanian fixed and mobile telephone network. | 28,356 | sentences | Low | nan | nan | nan | Arab | No | ELRA | With-Fee | 15,000.00€ | No | speech recognition | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
|
TRAD Arabic-French Mailing lists Parallel corpus - Test set | [] | nan | https://catalogue.elra.info/en-us/repository/browse/ELRA-W0105/ | Non Commercial Use - ELRA END USER | 2,016 | multilingual | ar-MSA: (Arabic (Modern Standard Arabic)) | other | text | crawling | This is a parallel corpus of 10,000 words in Arabic and 4 reference translations in French. The source texts are emails collected from Wikiar-I, a mailing list for discussions about the Arabic Wikipedia. | 10,000 | tokens | Low | nan | nan | nan | nan | Arab | No | ELRA | With-Fee | 300.00€ | No | machine translation | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
TRAD Arabic-French Mailing lists Parallel corpus - Development set | [] | nan | https://catalogue.elra.info/en-us/repository/browse/ELRA-W0107/ | Non Commercial Use - ELRA END USER | 2,016 | multilingual | ar-MSA: (Arabic (Modern Standard Arabic)) | other | text | crawling | This is a parallel corpus of 10,000 words in Arabic and a reference translation in French. The source texts are emails collected from Wikiar-I, a mailing list for discussions about the Arabic Wikipedia. | 10,000 | tokens | Low | nan | nan | nan | nan | Arab | No | ELRA | With-Fee | 300.00€ | No | machine translation | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
TRAD Arabic-English Web domain (blogs) Parallel corpus | [] | nan | https://catalogue.elra.info/en-us/repository/browse/ELRA-W0104/ | Non Commercial Use - ELRA END USER | 2,016 | multilingual | ar-MSA: (Arabic (Modern Standard Arabic)) | other | text | other | This is a parallel corpus of 10,000 words in Arabic and 2 reference translations in English. The source texts are blog articles written between 2008 and 2013. The translation has been conducted by two different translation teams following a strict protocol aimed at producing high quality translations. | 10,000 | tokens | Low | nan | nan | nan | nan | Arab | No | ELRA | With-Fee | 300.00€ | No | machine translation | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
opus_ubuntu | [] | https://huggingface.co/datasets/opus_ubuntu | https://huggingface.co/datasets/opus_ubuntu | BSD | 2,012 | multilingual | ar-CLS: (Arabic (Classic)) | other | text | crawling and annotation(translation) | These are translations of the Ubuntu software package messages, donated by the Ubuntu community. | 299 | documents | Low | OPUS | nan | Parallel Data, Tools and Interfaces in OPUS | http://www.lrec-conf.org/proceedings/lrec2012/pdf/463_Paper.pdf | Arab | No | HuggingFace | Free | nan | No | translation | LREC | 1203.0 | conference | International Conference on Language Resources and Evaluation | Jorg Tiedemann | Uppsala University | This paper presents the current status of OPUS, a growing language resource of parallel corpora and related tools. The focus in OPUS is to provide freely available data sets in various formats together with basic annotation to be useful for applications in computational linguistics, translation studies and cross-linguistic corpus studies. In this paper, we report about new data sets and their features, additional annotation tools and models provided from the website and essential interfaces and on-line services included in the project. | Khalid N. Elmadani |
ALR: Arabic Laptop Reviews dataset | [] | nan | https://github.com/bashartalafha/Arabic-Laptop-Reviews-ALR-Dataset | unknown | 2,017 | ar | mixed | reviews | text | crawling and annotation(other) | Arabic Laptops Reviews (ALR) dataset focuses on laptops reviews written in Arabic | 1,753 | sentences | Low | Jordan University of Science and Technology, Al-Balqa' Applied UniversityAl-Buraimi University College, Texas A&M University-San Antonio | nan | Aspect-Based Sentiment Analysis of Arabic Laptop Reviews | https://www.researchgate.net/publication/329557366_Aspect-Based_Sentiment_Analysis_of_Arabic_Laptop_Reviews | Arab | No | GitHub | Free | nan | No | sentiment analysis | ACIT | 1.0 | conference | international Arab Conference on Information Technology | Mahmoud Al-Ayyoub, Amal Gigieh, Areej Al-Qwaqenah, Mohammed N. Al-Kabi, Bashar Talafhah, Izzat Alsmadi | Jordan University of Science and Technology, Al-Balqa' Applied UniversityAl-Buraimi University College, Texas A&M University-San Antonio | Sentiment Analysis (SA) is one of the hottest research areas in Natural Language Processing (NLP) with vast commercial as well as academic applications. One of the most interesting versions of SA is called Aspect-Based SA (ABSA). Currently, most of the researchers focus on English text. Other languages such as Arabic have received less attention. To the best of our knowledge, only few papers have addressed ABSA of Arabic reviews and they have all been applied on only three datasets. In this work, we demonstrate our efforts to build the Arabic Laptops Reviews (ALR) dataset, which focuses on laptops reviews written in Arabic. To make it easy to use, the ALR dataset is prepared according to the annotation scheme of SemEval16-Task5. The annotation scheme considers two problems: aspect category prediction and sentiment polarity label prediction. It also comes with an evaluation procedure that extracts n-grams’ features and employs a Support Vector Machine (SVM) classifier in order to allow researchers to gauge and compare the performance of their systems. The evaluation results show that there is a lot of room for improvements in the performance of the SVM classifier for the aspect category prediction problem. As for the sentiment polarity label prediction, SVM’s accuracy is actually high. | Wafaa Mohammed |
Twt15DA_Lists | [
{
"Name": "Yemeni",
"Dialect": "ar-YE: (Arabic (Yemen))",
"Volume": "20,004",
"Unit": "sentences"
},
{
"Name": "Omani",
"Dialect": "ar-OM: (Arabic (Oman))",
"Volume": "20,861",
"Unit": "sentences"
},
{
"Name": "Saudi",
"Dialect": "ar-SA: (Arabic (Saudi Arabia))",
"Volume": "21,110",
"Unit": "sentences"
},
{
"Name": "Emirati",
"Dialect": "ar-AE: (Arabic (United Arab Emirates))",
"Volume": "20,957",
"Unit": "sentences"
},
{
"Name": "Qatari",
"Dialect": "ar-QA: (Arabic (Qatar))",
"Volume": "22,160",
"Unit": "sentences"
},
{
"Name": "Bahraini",
"Dialect": "ar-BH: (Arabic (Bahrain))",
"Volume": "22,160",
"Unit": "sentences"
},
{
"Name": "Kuwaiti",
"Dialect": "ar-KW: (Arabic (Kuwait))",
"Volume": "20,338",
"Unit": "sentences"
},
{
"Name": "Iraqi",
"Dialect": "ar-IQ: (Arabic (Iraq))",
"Volume": "20,241",
"Unit": "sentences"
},
{
"Name": "Jordanian",
"Dialect": "ar-JO: (Arabic (Jordan))",
"Volume": "19,762",
"Unit": "sentences"
},
{
"Name": "Syrian",
"Dialect": "ar-SY: (Arabic (Syria))",
"Volume": "18,750",
"Unit": "sentences"
},
{
"Name": "Egyptian",
"Dialect": "ar-EG: (Arabic (Egypt))",
"Volume": "20,109",
"Unit": "sentences"
},
{
"Name": "Libyan",
"Dialect": "ar-LY: (Arabic (Libya))",
"Volume": "22,844",
"Unit": "sentences"
},
{
"Name": "Tunisian",
"Dialect": "ar-TN: (Arabic (Tunisia))",
"Volume": "21,440",
"Unit": "sentences"
},
{
"Name": "Algerian",
"Dialect": "ar-DZ: (Arabic (Algeria))",
"Volume": "21,358",
"Unit": "sentences"
},
{
"Name": "Moroccan",
"Dialect": "ar-MA: (Arabic (Morocco))",
"Volume": "20,735",
"Unit": "sentences"
}
] | https://huggingface.co/datasets/arbml/Twt15DA_Lists | https://github.com/Maha-J-Althobaiti/Twt15DA_Lists | CC BY-NC-ND 4.0 | 2,021 | ar | mixed | social media | text | crawling and annotation(translation) | The annotated dialectal Arabic corpus (Twt15DA) is collected from Twitter and consists of 311,785 tweets containing 3,858,459 words in total. They randomly selected a sample of 75 tweets per country, 1125 tweets in total, and conducted a manual dialect identification task by native speakers. | 311,785 | sentences | Low | Taif University | nan | Creation of annotated country-level dialectal Arabic resources: An unsupervised approach | https://web.archive.org/web/20210813220628id_/https://www.cambridge.org/core/services/aop-cambridge-core/content/view/2DE64B777EF0277557AFA90E2BB75B62/S135132492100019Xa.pdf/div-class-title-creation-of-annotated-country-level-dialectal-arabic-resources-an-unsupervised-approach-div.pdf | Arab | No | GitHub | Free | nan | No | dialect identification | Cambridge University Press | 1.0 | journal | Natural Language Engineering (2021), Cambridge University Press | Maha J. Althobaiti | Department of Computer Science, College of Computers and Information Technology, Taif University | The wide usage of multiple spoken Arabic dialects on social networking sites stimulates increasing interest in Natural Language Processing (NLP) for dialectal Arabic (DA). Arabic dialects represent true linguistic diversity and differ from modern standard Arabic (MSA). In fact, the complexity and variety of these dialects make it insufficient to build one NLP system that is suitable for all of them. In comparison with MSA, the available datasets for various dialects are generally limited in terms of size, genre and scope. In this article, we present a novel approach that automatically develops an annotated country-level dialectal Arabic corpus and builds lists of words that encompass 15 Arabic dialects. The algorithm uses an iterative procedure consisting of two main components: automatic creation of lists for dialectal words and automatic creation of annotated Arabic dialect identification corpus. To our knowledge, our study is the first of its kind to examine and analyse the poor performance of the MSA part-of-speech tagger on dialectal Arabic contents and to exploit that in order to extract the dialectal words. The pointwise mutual information association measure and the geographical frequency of word occurrence online are used to classify dialectal words. The annotated dialectal Arabic corpus (Twt15DA), built using our algorithm, is collected from Twitter and consists of 311,785 tweets containing 3,858,459 words in total. We randomly selected a sample of 75 tweets per country, 1125 tweets in total, and conducted a manual dialect identification task by native speakers. The results show an average inter-annotator agreement score equal to 64%, which reflects satisfactory agreement considering the overlapping features of the 15 Arabic dialects. | Mustafa Ghaleb |
TRAD Arabic-English Parallel corpus of transcribed Broadcast News Speech | [] | nan | https://catalogue.elra.info/en-us/repository/browse/ELRA-W0102/ | Non Commercial Use - ELRA END USER | 2,016 | multilingual | ar-MSA: (Arabic (Modern Standard Arabic)) | transcribed audio | text | other | This is a parallel corpus of 10,000 words in Arabic and 2 reference translations in English. The source texts are transcriptions of broadcast news in Arabic recorded on France 24. The translation has been conducted by two different translation teams following a strict protocol aimed at producing high quality translations. | 10,000 | tokens | Low | nan | nan | nan | nan | Arab | No | ELRA | With-Fee | 300.00€ | No | machine translation, speech recognition | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
Arabic-Multi-Classification-Dataset-AMCD | [] | https://huggingface.co/datasets/arbml/AMCD | https://github.com/waelyafooz/Arabic-Multi-Classification-Dataset-AMCD | unknown | 2,021 | ar | mixed | commentary | text | crawling | This is an Arabic dataset called Arabic Multi Classification Dataset version 0.1 (AMCD). AMCD can be used for texting mining and clustering and classification algorithm. It collected from YouTube Videos meta data and user comments. It can be used for topic modelling, text summarization, apply classification or clustering algorithms. | 8,046 | sentences | Low | nan | nan | nan | nan | Arab | No | GitLab | Free | nan | No | topic modeling, summarization | nan | nan | nan | nan | nan | Waelyafooz | nan | Abdelrahman Rezk |
Arabic Treebank: Part 3 | [] | nan | https://catalog.ldc.upenn.edu/LDC2005T20 | LDC User Agreement | 2,005 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | news articles | text | manual curation | This dataset is a collection of 300,00 Arabic tokens with their syntactic treebank annotation and annotation on part of speech (POS), gloss, and word segmentation. | 300,000 | tokens | Low | LDC | part of the dataset was derived from Arabic Gigaword (LDC2003T12) | nan | nan | Arab | No | LDC | With-Fee | 3,500.00 $ | No | information retrieval, cross-lingual information retrieval, information detection, | nan | 24.0 | nan | nan | Mohamed Maamouri, Ann Bies, Tim Buckwalter, Hubert Jin, Wigdan Mekki | nan | nan | Maged S. Alshaibani |
Arabic 100k Reviews | [] | nan | https://www.kaggle.com/datasets/abedkhooli/arabic-100k-reviews | unknown | 2,022 | ar | mixed | reviews | text | other | Few Arabic datasets are available for classification comparison and other NLP tasks. This dataset is mainly a compilation of several available datasets and a sampling of 100k rows (99999 to be exact). | 99,999 | documents | Low | Abed Khooli | The hotels and book reviews are a subset of [HARD](HARD: https://github.com/elnagara/HARD-Arabic-Dataset ) and BRAD. The rest were selected from hadyelsahar with a little over 100 airlines reviews collected manually. | nan | nan | Arab | No | kaggle | Free | nan | No | sentiment analysis | nan | nan | nan | nan | nan | nan | nan | Afrah Altamimi |
TRAD Arabic-English Newspaper Parallel corpus - Test set 1 | [] | nan | https://catalogue.elra.info/en-us/repository/browse/ELRA-W0099/ | Non Commercial Use - ELRA END USER | 2,016 | multilingual | ar-MSA: (Arabic (Modern Standard Arabic)) | other | text | crawling and annotation(translation) | This is a parallel corpus of 10,000 words in Arabic and 2 reference translations in English. The source texts are articles collected in 2012 from the Arabic version of Le Monde Diplomatique. The translation has been conducted by two different translation teams following a strict protocol aimed at producing high quality translations. | 10,000 | tokens | Low | nan | nan | nan | nan | Arab | No | ELRA | With-Fee | 300.00€ | No | machine translation | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
LAMA | [] | https://huggingface.co/datasets/arbml/ara_emotion | https://github.com/UBC-NLP/ara_emotion_naacl2018 | unknown | 2,018 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | social media | text | crawling and annotation(translation) | A dataset for Modern Standard and Dialectal Arabic emotion detection focused at Robert Plutchik’s 8 basic emotion types | 7,000 | sentences | Medium | University of British Columbia | nan | Enabling Deep Learning of Emotion With First-Person Seed Expressions | https://aclanthology.org/W18-1104.pdf | Arab | No | GitHub | Free | nan | Yes | emotion detection | ACL | 21.0 | workshop | he Second Workshop on Computational Modeling of People’s Opinions, Personality, and Emotions in Social Media, ACL | Hassan Alhuzali, Muhammad Abdul-Mageed, Lyle Ungar | nan | The computational treatment of emotion in natural language text remains relatively lim- ited, and Arabic is no exception. This is partly due to lack of labeled data. In this work, we describe and manually validate a method for the automatic acquisition of emotion labeled data and introduce a newly developed data set for Modern Standard and Dialectal Arabic emotion detection focused at Robert Plutchik’s 8 basic emotion types. Using a hybrid supervi- sion method that exploits first person emotion seeds, we show how we can acquire promis- ing results with a deep gated recurrent neu- ral network. Our best model reaches 70% F- score, significantly (i.e., 11%, p < 0.05) out- performing a competitive baseline. Applying our method and data on an external dataset of 4 emotions released around the same time we fi- nalized our work, we acquire 7% absolute gain in F-score over a linear SVM classifier trained on gold data, thus validating our approach | Emad A. Alghamdi |
TRAD Arabic-English Mailing lists Parallel corpus - Test set | [] | nan | https://catalogue.elra.info/en-us/repository/browse/ELRA-W0106/ | Non Commercial Use - ELRA END USER | 2,016 | multilingual | ar-MSA: (Arabic (Modern Standard Arabic)) | wikipedia | text | crawling and annotation(translation) | This is a parallel corpus of 10,000 words in Arabic and 2 reference translations in English. The source texts are emails collected from Wikiar-I, a mailing list for discussions about the Arabic Wikipedia. The collected emails are dated from 2010 to 2012. The translation has been conducted by two different translation teams following a strict protocol aimed at producing high quality translations. | 10,000 | tokens | Low | nan | nan | nan | nan | Arab | No | ELRA | With-Fee | 300.00€ | No | machine translation | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
Quran Hadith Datasets | [] | https://huggingface.co/datasets/arbml/Quran_Hadith | https://github.com/ShathaTm/Quran_Hadith_Datasets | unknown | 2,022 | multilingual | ar-CLS: (Arabic (Classic)) | other | text | crawling and annotation(translation) | The datasets showcase the related and non-related pairs of Quran-Quran and Quran-Hadith. It has Classical Arabic and English translated verses and teachings. | 20,360 | sentences | Low | nan | nan | Challenging the Transformer-based models with a Classical Arabic dataset: Quran and Hadith | nan | Arab-Latn | No | GitHub | Free | nan | Yes | semantic similarity | nan | nan | conference | Language Resources and Evaluation Conference | Shatha Altammami, Eric Atwell | University of Leeds/King Saud University | nan | Abdullah Alsaleh |
xcsr | [
{
"Name": "X-CSQA",
"Dialect": "ar-CLS: (Arabic (Classic))",
"Volume": "2,074",
"Unit": "sentences"
},
{
"Name": "X-CODAH:",
"Dialect": "ar-CLS: (Arabic (Classic))",
"Volume": "1,300",
"Unit": "sentences"
}
] | https://huggingface.co/datasets/xcsr | https://huggingface.co/datasets/xcsr | unknown | 2,021 | multilingual | ar-CLS: (Arabic (Classic)) | other | text | human translation | To evaluate multi-lingual language models (ML-LMs) for commonsense reasoning in a cross-lingual zero-shot transfer setting (X-CSR) | 3,374 | sentences | Low | University of Southern California | nan | Common Sense Beyond English: Evaluating and Improving Multilingual Language Models for Commonsense Reasoning | https://aclanthology.org/2021.acl-long.102.pdf | Arab | No | HuggingFace | Free | nan | Yes | commonsense reasoning | ACL | 13.0 | conference | Associations of computation linguistics | Bill Yuchen Lin, Seyeon Lee, Xiaoyang Qiao, Xiang Ren | University of Southern California | Commonsense reasoning research has so far been limited to English. We aim to evaluate and improve popular multilingual language models (ML-LMs) to help advance commonsense reasoning (CSR) beyond English. We collect the Mickey corpus, consisting of 561k sentences in 11 different languages, which can be used for analyzing and improving ML-LMs. We propose Mickey Probe, a language-general probing task for fairly evaluating the common sense of popular ML-LMs across different languages. In addition, we also create two new datasets, X-CSQA and X-CODAH, by translating their English versions to 14 other languages, so that we can evaluate popular ML-LMs for cross-lingual commonsense reasoning. To improve the performance beyond English, we propose a simple yet effective method — multilingual contrastive pretraining (MCP). It significantly enhances sentence representations, yielding a large performance gain on both benchmarks (e.g., +2.7% accuracy for X-CSQA over XLM-R_L). | Khalid N. Elmadani |
TRAD Arabic-English Mailing lists Parallel corpus - Development set | [] | nan | https://catalogue.elra.info/en-us/repository/browse/ELRA-W0108/ | Non Commercial Use - ELRA END USER | 2,016 | multilingual | ar-MSA: (Arabic (Modern Standard Arabic)) | wikipedia | text | crawling and annotation(translation) | This is a parallel corpus of 10,000 words in Arabic and a reference translation in English. The source texts are emails collected from Wikiar-I, a mailing list for discussions about the Arabic Wikipedia. The collected emails are dated from 2004 to 2007. The translation has been conducted following a strict protocol aimed at producing high quality translations. | 10,000 | tokens | Low | nan | nan | nan | nan | Arab | No | ELRA | With-Fee | nan | No | machine translation | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
OrienTel Tunisia MSA (Modern Standard Arabic) database | [] | nan | https://catalogue.elra.info/en-us/repository/browse/ELRA-S0187/ | Non Commercial Use - ELRA END USER | 2,005 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | other | spoken | other | The OrienTel Tunisia MSA (Modern Standard Arabic) database comprises 598 Tunisian speakers (359 males, 239 females) recorded over the Tunisian fixed and mobile telephone network. | 31,096 | sentences | Low | nan | nan | nan | nan | Arab | No | ELRA | With-Fee | 6,000.00€ | No | speech recognition | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
ASKFM | [] | https://huggingface.co/datasets/arbml/ASKFM | https://github.com/Omarito2412/ASKFM | MIT License | 2,017 | ar | mixed | web pages | text | crawling | This dataset is a merge of 98k questions and their respective answers as written by different authors on Askfm. | 98,000 | sentences | High | nan | nan | nan | nan | Arab | No | GitHub | Free | nan | No | information retrieval, question answering | nan | nan | nan | nan | Omar Essam | nan | nan | Mustafa Ghaleb |
ml_spoken_words | [] | https://huggingface.co/datasets/MLCommons/ml_spoken_words | https://mlcommons.org/en/multilingual-spoken-words/ | CC BY 4.0 | 2,021 | multilingual | ar-CLS: (Arabic (Classic)) | transcribed audio | spoken | other | Multilingual Spoken Words Corpus (MSWC), a large and growing audio dataset of spoken words in 50 different languages. | 7.6 | hours | Low | Coqui, Factored, Google, Harvard University, Intel, Landing AI, NVIDIA, University of Michigan | Common Voice dataset | Multilingual Spoken Words Corpus | https://datasets-benchmarks-proceedings.neurips.cc/paper/2021/file/fe131d7f5a6b38b23cc967316c13dae2-Paper-round2.pdf | Arab | No | HuggingFace | Free | nan | Yes | keyword spotting, spoken term search | NeurIPS | 1.0 | conference | The Conference on Neural Information Processing Systems | Mark Mazumder, Sharad Chitlangia, Colby Banbury, Yiping Kang, Juan Ciro, Keith Achorn, Daniel Galvez, Mark Sabini, Peter Mattson, David Kanter, Greg Diamos, Pete Warden, Josh Meyer, Vijay Janapa Reddi | nan | Coqui, Factored, Google, Harvard University, Intel, Landing AI, NVIDIA, University of Michigan | Wafaa Mohammed |
Stopword Lists for 19 Languages | [] | nan | https://www.kaggle.com/datasets/rtatman/stopword-lists-for-19-languages/download | unknown | 2,017 | multilingual | ar-MSA: (Arabic (Modern Standard Arabic)) | other | text | other | Some words, like “the” or “and” in English, are used a lot in speech and writing. For most Natural Language Processing applications, you will want to remove these very frequent words. This is usually done using a list of “stopwords” which has been complied by hand. | 19 | documents | Low | nan | nan | nan | nan | Arab | No | kaggle | Free | nan | No | stop words | nan | nan | nan | nan | nan | nan | nan | Afrah Altamimi |
bible_para | [] | https://huggingface.co/datasets/bible_para | https://huggingface.co/datasets/bible_para | CC0 | 2,014 | multilingual | ar-CLS: (Arabic (Classic)) | books | text | human translation | This is a multilingual parallel corpus created from translations of the Bible | 2,800,000 | tokens | Low | OPUS | nan | A massively parallel corpus: the Bible in 100 languages | https://link.springer.com/content/pdf/10.1007/s10579-014-9287-y.pdf | Arab | No | HuggingFace | Free | nan | No | machine translation | Language Resources and Evaluation | 49.0 | journal | Language Resources and Evaluation | Christos Christodoulopoulos, Mark Steedman | UIUC, University of Edinburgh | We describe the creation of a massively parallel corpus based on 100 translations of the Bible. We discuss some of the difficulties in acquiring and processing the raw material as well as the potential of the Bible as a corpus for natural language processing. Finally we present a statistical analysis of the corpora collected and a detailed comparison between the English translation and other English corpora. | Khalid N. Elmadani |
Arabic business corpora | [] | https://huggingface.co/datasets/arbml/buisness_corpora | https://sourceforge.net/projects/arabic-business-copora/ | unknown | 2,016 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | web pages | text | crawling | The main corpora contains 1200 articles. The articles have been tagged using Stanford Arabic Part of Speech Tagger. | 1,200 | documents | Low | nan | nan | nan | nan | Arab | No | sourceforge | Free | nan | No | part of speech tagging, topic classification | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
OSMAN UN Corpus | [] | nan | https://github.com/drelhaj/OsmanReadability | unknown | 2,016 | multilingual | ar-MSA: (Arabic (Modern Standard Arabic)) | commentary | text | other | Arabic readability | 73,000 | sentences | Medium | Lancaster University | nan | OSMAN – A Novel Arabic Readability Metric | https://aclanthology.org/L16-1038.pdf | Arab-Latn | No | GitHub | Free | nan | No | readability assessnebty | European Language Resources Association (ELRA) | nan | conference | European Language Resources Association (ELRA) | Mahmoud El-Ha, Paul Rayson | Lancaster University | We present OSMAN (Open Source Metric for Measuring Arabic Narratives) - a novel open source Arabic readability metric and tool. It allows researchers to calculate readability for Arabic text with and without diacritics. OSMAN is a modified version of the conventional readability formulas such as Flesch and Fog. In our work we introduce a novel approach towards counting short, long and stress syllables in Arabic which is essential for judging readability of Arabic narratives. We also introduce an additional factor called “Faseeh” which considers aspects of script usually dropped in informal Arabic writing. To evaluate our methods we used Spearman’s correlation metric to compare text readability for 73,000 parallel sentences from English and Arabic UN documents. The Arabic sentences were written with the absence of diacritics and in order to count the number of syllables we added the diacritics in using an open source tool called Mishkal. The results show that OSMAN readability formula correlates well with the English ones making it a useful tool for researchers and educators working with Arabic tex | Emad A. Alghamdi |
International Corpus of Arabic | [] | nan | http://www.bibalex.org/ica/ar/ | unknown | 2,007 | ar | mixed | other | text | other | There are two points of view about the need for corpora. The first one says that there can not be any corpora, however large, that contain information about all of the areas of any language lexicon and grammar of that language. The second point of view is that every corpus, however small, has taught the person facts that could not be imagined finding out about in any other way. | nan | sentences | Low | Alexandria University | nan | Building an International Corpus of Arabic (ICA): Progress of Compilation Stage | http://www.bibalex.org/isis/UploadedFiles/Publications/Building%20an%20Intl%20corpus%20of%20arabic.pdf | Arab | No | other | Upon-Request | nan | No | text generation, language modeling | nan | 35.0 | conference | 7th international conference on language engineering, Cairo, Egypt | Sameh Alansary, Magdy Nagi, Noha Adly | nan | This paper focuses on three axes. The first ax is gives a survey of the importance of corpora in language studies e.g. lexicography, grammar, sem antics, Natural Language Processing and other areas. The second axis demonstrates how the A rabic language lacks textual resources, such as corpora and tools for corpus analysis and t he effected of this lack on the quality of Arabic language applications. There are rarely succ essful trials in compiling Arabic corpora, therefore, the third axis presents the technical de sign of the International Corpus of Arabic (ICA), a newly established representative corpus of Arabic that is intended to cover the Arabic language as being used all over the Arab world. The corpus is planned to support various Arabic studies that depends on authentic data, in a ddition to building Arabic Natural Language Processing Applications. | Abdelrahman Rezk |
un_multi | [] | https://huggingface.co/datasets/un_multi | https://huggingface.co/datasets/un_multi | unknown | 2,010 | multilingual | ar-CLS: (Arabic (Classic)) | other | text | human translation | This is a collection of translated documents from the United Nations. | 300,000,000 | tokens | Low | UN | nan | MultiUN: A Multilingual Corpus from United Nation Documents | http://www.lrec-conf.org/proceedings/lrec2010/pdf/686_Paper.pdf | Arab | No | HuggingFace | Free | nan | No | machine translation | LREC | 228.0 | conference | International Conference on Language Resources and Evaluation | Andreas Eisele, Yu Chen | German Research Center for Artificial Intelligence (DFKI) | This paper describes the acquisition, preparation and properties of a corpus extracted from the official documents of the United Nations (UN). This corpus is available in all 6 official languages of the UN, consisting of around 300 million words per language. We describe the methods we used for crawling, document formatting, and sentence alignment. This corpus also includes a common test set for machine translation. We present the results of a French-Chinese machine translation experiment performed on this corpus. | Khalid N. Elmadani |
CQA-MD: SemEval-2016 Task 3 | [] | https://huggingface.co/datasets/arbml/CQA_MD_ar | https://alt.qcri.org/semeval2016/task3/index.php?id=data-and-tools | unknown | 2,016 | multilingual | ar-MSA: (Arabic (Modern Standard Arabic)) | web pages | text | crawling and annotation(other) | It includes a TRAIN/DEV split with reliable double-checked DEV (1,281 original questions, and 37,795 potentially related question-answer pairs) + unannotated (163,383 question--answer pairs) | 45,164 | sentences | Low | QCRI | nan | SemEval-2016 Task 3: Community Question Answering | https://alt.qcri.org/semeval2016/task3/data/uploads/semeval2016-task3-report.pdf | Arab | No | QCRI Resources | Free | nan | Yes | question answering | semEval | nan | conference | International Workshop on Semantic Evaluation | Preslav Nakov, Llu´ıs Marquez, Alessandro Moschitti, `
Walid Magdy, Hamdy Mubarak, Abed Alhakim Freihat | ALT Research Group, Qatar Computing Research Institute, HBKU | This paper describes the SemEval–2016
Task 3 on Community Question Answering, which we offered in English and Arabic. For English, we had three subtasks: Question–Comment Similarity (subtask
A), Question–Question Similarity (B), and
Question–External Comment Similarity (C).
For Arabic, we had another subtask: Rerank
the correct answers for a new question (D).
Eighteen teams participated in the task, submitting a total of 95 runs (38 primary and 57
contrastive) for the four subtasks. A variety
of approaches and features were used by the
participating systems to address the different
subtasks, which are summarized in this paper.
The best systems achieved an official score
(MAP) of 79.19, 76.70, 55.41, and 45.83 in
subtasks A, B, C, and D, respectively. These
scores are significantly better than those for
the baselines that we provided. For subtask A,
the best system improved over the 2015 winner by 3 points absolute in terms of Accuracy | Zaid Alyafeai |
xquad_r | [] | https://huggingface.co/datasets/xquad_r | https://github.com/google-research-datasets/lareqa | CC BY 4.0 | 2,020 | multilingual | mixed | other | text | machine translation | XQuAD-R is a retrieval version of the XQuAD dataset (a cross-lingual extractive QA dataset). Like XQuAD, XQUAD-R is an 11-way parallel dataset, where each question appears in 11 different languages and has 11 parallel correct answers across the languages | 1,190 | sentences | Low | Google Research | XQuAD dataset | LAReQA: Language-agnostic answer retrieval from a multilingual pool | https://arxiv.org/pdf/2004.05484.pdf | Arab | No | GitHub | Free | nan | No | language-agnostic answer retrieval from a multilingual candidate pool | nan | 21.0 | nan | nan | Uma Roy, Noah Constant, Rami Al-Rfou, Aditya Barua, Aaron Phillips, Yinfei Yang | Google Research | We present LAReQA, a challenging new benchmark for language-agnostic answer retrieval from a multilingual candidate pool. Unlike previous cross-lingual tasks, LAReQA tests for “strong” cross-lingual alignment, requiring semantically related cross-language pairs to be closer in representation space than unrelated same-language pairs. Building on multilingual BERT (mBERT), we study different strategies for achieving strong alignment. We find that augmenting training data via machine translation is effective, and improves significantly over using mBERT out-of-the-box. Interestingly, the embedding baseline that performs the best on LAReQA falls short of competing baselines on zero-shot variants of our task that only target “weak” alignment. This finding underscores our claim that language agnostic retrieval is a substantively new kind of cross-lingual evaluation. | Wafaa Mohammed |
xlel_wd_dictionary | [] | https://huggingface.co/datasets/adithya7/xlel_wd_dictionary | https://huggingface.co/datasets/adithya7/xlel_wd_dictionary | CC BY 4.0 | 2,022 | multilingual | ar-CLS: (Arabic (Classic)) | wikipedia | text | crawling | XLEL-WD is a multilingual event linking dataset. This supplementary dataset contains a dictionary of event items from Wikidata. The descriptions for Wikidata event items are taken from the corresponding multilingual Wikipedia articles. | 114,834 | sentences | Low | Carnegie Mellon University | nan | Multilingual Event Linking to Wikidata | https://arxiv.org/pdf/2204.06535.pdf | Arab | No | HuggingFace | Free | nan | No | event linking | nan | nan | nan | nan | Adithya Pratapa, Rishubh Gupta, Teruko Mitamura | Carnegie Mellon University | We present a task of multilingual linking of events to a knowledge base. We automatically compile a large-scale dataset for this task, comprising of 1.8M mentions across 44 languages referring to over 10.9K events from Wikidata. We propose two variants of the event linking task: 1) multilingual, where event descriptions are from the same language as the mention, and 2) crosslingual, where all event descriptions are in English. On the two proposed tasks, we compare multiple event linking systems including BM25+ (Lv and Zhai, 2011) and multilingual adaptations of the biencoder and crossencoder architectures from BLINK (Wu et al., 2020). In our experiments on the two task variants, we find both biencoder and crossencoder models significantly outperform the BM25+ baseline. Our results also indicate that the crosslingual task is in general more challenging than the multilingual task. To test the out-of-domain generalization of the proposed linking systems, we additionally create a Wikinews-based evaluation set. We present qualitative analysis highlighting various aspects captured by the proposed dataset, including the need for temporal reasoning over context and tackling diverse event descriptions across languages. | Khalid N. Elmadani |
Annotated Corpus of Arabic Al-Quran Question and Answer | [] | nan | https://archive.researchdata.leeds.ac.uk/464/1/AAQQAC.XML | CC BY 4.0 | 2,018 | ar | mixed | other | text | other | QQAC is a collection of approximately 2224 questions and answers about Al-Al-Quran. Each question and answer is annotated with the question ID, question word (particles), chapter number, verse number, question topic, question type, Al-Quran ontology concepts (Alqahtani & Atwell, 2018) and question source. The aim of this corpus is to provide a Question-Answering taxonomy for questions about Al-Quran. | 1,224 | sentences | High | University of Leeds | nan | Annotated Corpus of Arabic Al-Quran Question and Answer | https://archive.researchdata.leeds.ac.uk/464/ | Arab | No | other | Free | nan | No | question answering | nan | nan | preprint | nan | Alqahtani, Mohammad and Atwell | nan | AQQAC is a collection of approximately 2224 questions and answers about Al-Al-Quran. Each question and answer is annotated with the question ID, question word (particles), chapter number, verse number, question topic, question type, Al-Quran ontology concepts (Alqahtani & Atwell, 2018) and question source. The aim of this corpus is to provide a Question-Answering taxonomy for questions about Al-Quran. Additionally, this corpus might be used as a data set for testing and evaluating Islamic IR systems. The text of Al-Quran questions and answers were extracted from trusted two islamic sources: (1000 Su'al Wa Jawab Fi ALKORAN) was compiled by the famous Islamic scholar Ashur (2001). This book contains 1000 questions and answers about Al-Quran written in the Arabic language. Islam – Al-Quran and Tafseer is a website about Al-Quran that includes a description and a translation of Al-Quran and the reciting rules, the “Tajweed”. Additionally, this website has approximately 1224 questions and answers about Al-Quran in the Arabic language extracted from the Altabari Tafseer. Currently, this dataset contains 1224 annotated question-answers and the missing data that hasn’t been shared is due to copyright concerns.
| Abdelrahman Rezk |
Corpus of Early Poetic Arabic (CEAP) | [] | https://huggingface.co/datasets/arbml/CEAP | https://sourceforge.net/projects/ceap-bp/ | unknown | 2,021 | ar | ar-CLS: (Arabic (Classic)) | web pages | text | crawling | It contains 50 TXT files recording poetry composition by various authors from 6th and 7th c. It was derived from two corpora: King Saud University Classical Arabic Corpus (KSUCAC) created by the team led by Maha S. Alrabiah (2014), and a corpus prepared by Abeer Alsheddi (2016). | 50 | documents | Low | nan | Alrabiah, Maha S. (2014): King Saud University Classical Arabic Corpus, Ar-Riyāḍ. Alsheddi, Abeer. (2016): Edit Distance Adapted to Natural Language Words. M.A. Thesis. Ar-Riyāḍ. | Cultural Conceptualizations of shame & dishonor in Early Poetic Arabic (EPA) | https://www.ejournals.eu/pliki/art/20227/ | Latn | No | sourceforge | Free | nan | No | cultural conceptualizations of shame and dishonor | The Polish Journal of the Arts and Culture | nan | journal | The Polish Journal of the Arts and Culture. New Series} | Bartosz Pietrzak | Institute of Oriental Studies of Jagiellonian University in Krakow. | Persisting in a binary relationship with honor, shame was an important element of the pre-Islamic Arabic social evaluation system. In my
study, I analyzed the two most important EPA concepts parallel to
English shame – ˁayb and ˁār – applying the Cultural Linguistic approach. Based on the analyses on corpus of Early Arabic poetry and
Classical Arabic dictionaries, I represented cultural schemata encoding the knowledge shared by pre-Islamic Arabs about those phenomena. The paper presents also metaphoric, metonymic, and image-schematic models, which account for the specifics of associated linguistic
frames. Moreover, I posit a hypothesis on the existence of a schema
subsuming the honor- and shame-dishonor-related schemata in
form of social evaluation of usefulness, which seems to correspond
to the historical and linguistic data. | Mustafa Ghaleb |
Curras: an annotated corpus for the Palestinian Arabic dialect | [] | nan | https://portal.sina.birzeit.edu/curras/download.html | custom | 2,016 | ar | ar-PS: (Arabic (Palestine)) | other | text | manual curation | Curras: a dataset for Palestinian Arabic with rich metadata including POS tagging, lemma, stem, and other | 56,700 | tokens | Medium | birzeit university | nan | Curras: an annotated corpus for the Palestinian Arabic dialect | http://www.jarrar.info/publications/JHRAZ17.pdf | Arab | No | other | Upon-Request | nan | No | named entity recognition, stemming, lemmatization | LREC | 76.0 | conference | Language Resource & Evaluation | Mustafa Jarrar, Nizar Habash, Faeq Alrimawi, Diyam Akra, Nasser Zalmout | Birzeit University, NYU | In this article we present Curras, the first morphologically annotated
corpus of the Palestinian Arabic dialect. Palestinian Arabic is one of the many
primarily spoken dialects of the Arabic language. Arabic dialects are generally
under-resourced compared to Modern Standard Arabic, the primarily written and
official form of Arabic. We start in the article with a background description that
situates Palestinian Arabic linguistically and historically and compares it to Modern
Standard Arabic and Egyptian Arabic in terms of phonological, morphological,
orthographic, and lexical variations. We then describe the methodology we developed to collect Palestinian Arabic text to guarantee a variety of representative
domains and genres. We also discuss the annotation process we used, which
extended previous efforts for annotation guideline development, and utilized
existing automatic annotation solutions for Standard Arabic and Egyptian Arabic.
The annotation guidelines and annotation meta-data are described in detail. The
Curras Palestinian Arabic corpus consists of more than 56 K tokens, which are annotated with rich morphological and lexical features. The inter-annotator agreement results indicate a high degree of consistency. | Maged S. Alshaibani |
NADiA: News Articles Dataset in Arabic for Multi-Label Text Categorization | [] | https://huggingface.co/datasets/arbml/NADiA | https://data.mendeley.com/datasets/hhrb7phdyx/1 | CC BY 4.0 | 2,019 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | news articles | text | crawling | NADiA Dataset is the largest, to the best of our knowledge, source for Arabic textual data that can be used in any NLP related task such as text classification. We chose the abbreviation NADiA as it is a common Arabic name. The data was collected by scraping ‘SkyNewsArabia’ and ‘Masrawy’ news websites using Python scripts that are fine-tuned for each website. SkyNewsArabia will be referred to as NADiA1, while the latter would be NADiA2. NADiA1 is a big dataset containing 37,445 files, while NADiA2 is a huge dataset that contains 678,563 files. However, after filtering and cleaning we reduced the numbers to 35,416 and 451,230 for NADiA 1 and 2, respectively. | 486,646 | documents | nan | University of Sharjah | nan | nan | nan | Arab | No | Mendeley Data | Free | nan | No | multi-label text categorization | nan | nan | nan | nan | Al-Debsi Ridhwan,Elnagar Ashraf,Einea Omar | University of Sharjah | nan | Kamel GAANOUN |
ABSA-Hotels | [] | nan | https://github.com/msmadi/ABSA-Hotels | MIT License | 2,016 | multilingual | mixed | web pages | text | crawling and annotation(other) | Around 15,562 Hotels' reviews were thoroughly reviewed by this research authors and a subset of 2,291 reviews were selected. The original dataset has been collected from well known Hotels' booking websites such as Booking.com, TripAdvisor.com. | 24,028 | sentences | Low | nan | nan | SemEval-2016 Task 5: Aspect Based Sentiment Analysis | https://aclanthology.org/S16-1002.pdf | Arab | No | GitHub | Free | nan | Yes | review classification | SemEval | nan | conference | International Workshop on Semantic Evaluation | Maria Pontiki, Dimitrios Galanis, Haris Papageorgiou, Ion Androutsopoulos,Suresh Manandhar, Mohammad AL-Smadi, Mahmoud Al-Ayyoub, Yanyan Zhao,Bing Qin5, Orphée De Clercq, Véronique Hoste, Marianna Apidianaki,Xavier Tannier, Natalia Loukachevitch, Evgeny Kotelnikov,Nuria Bel, Salud María Jiménez-Zafra, Gülşen Eryiğit | nan | This paper describes the SemEval 2016 shared
task on Aspect Based Sentiment Analysis
(ABSA), a continuation of the respective tasks
of 2014 and 2015. In its third year, the task
provided 19 training and 20 testing datasets
for 8 languages and 7 domains, as well as a
common evaluation procedure. From these
datasets, 25 were for sentence-level and 14 for
text-level ABSA; the latter was introduced for
the first time as a subtask in SemEval. The task
attracted 245 submissions from 29 teams. | Zaid Alyafeai |
xlel_wd | [] | https://huggingface.co/datasets/adithya7/xlel_wd | https://huggingface.co/datasets/adithya7/xlel_wd | CC BY 4.0 | 2,022 | multilingual | ar-CLS: (Arabic (Classic)) | wikipedia | text | crawling | XLEL-WD is a multilingual event linking dataset. This dataset contains mention references in multilingual Wikipedia/Wikinews articles to event items from Wikidata. The descriptions for Wikidata event items are taken from the corresponding Wikipedia articles. | 10,947 | documents | Low | Carnegie Mellon University | nan | Multilingual Event Linking to Wikidata | https://arxiv.org/pdf/2204.06535.pdf | Arab | No | HuggingFace | Free | nan | Yes | multilingual linking, crosslingual linking | nan | nan | nan | nan | Adithya Pratapa, Rishubh Gupta, Teruko Mitamura | Carnegie Mellon University | We present a task of multilingual linking of events to a knowledge base. We automatically compile a large-scale dataset for this task, comprising of 1.8M mentions across 44 languages referring to over 10.9K events from Wikidata. We propose two variants of the event linking task: 1) multilingual, where event descriptions are from the same language as the mention, and 2) crosslingual, where all event descriptions are in English. On the two proposed tasks, we compare multiple event linking systems including BM25+ (Lv and Zhai, 2011) and multilingual adaptations of the biencoder and crossencoder architectures from BLINK (Wu et al., 2020). In our experiments on the two task variants, we find both biencoder and crossencoder models significantly outperform the BM25+ baseline. Our results also indicate that the crosslingual task is in general more challenging than the multilingual task. To test the out-of-domain generalization of the proposed linking systems, we additionally create a Wikinews-based evaluation set. We present qualitative analysis highlighting various aspects captured by the proposed dataset, including the need for temporal reasoning over context and tackling diverse event descriptions across languages. | Khalid N. Elmadani |
Al-Hayat Arabic Corpus | [] | nan | https://catalogue.elra.info/en-us/repository/browse/ELRA-W0030/ | Non Commercial Use - ELRA END USER | 2,002 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | news articles | text | crawling | The corpus was developed in the course of a research project at the University of Essex, in collaboration with the Open University. The corpus contains Al-Hayat newspaper articles with value added for Language Engineering and Information Retrieval applications development purposes. The data have been distributed into 7 subject-specific databases, thus following the Al-Hayat subject tags: General, Car, Computer, News, Economics, Science, and Sport. Mark-up, numbers, special characters and punctuation have been removed. The size of the total file is 268 MB. The dataset contains 18,639,264 distinct tokens in 42,591 articles, organised in 7 domains | 42,591 | documents | Low | nan | nan | nan | nan | Arab | No | ELRA | With-Fee | 720.00€ | No | topic classification | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
Arabic WikiReading and KaifLematha | [] | nan | https://github.com/esulaiman/Arabic-WikiReading-and-KaifLematha-datasets | unknown | 2,022 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | wikipedia | text | crawling and annotation(other) | high quality and large-scale Arabic reading comprehension datasets: Arabic WikiReading and KaifLematha with around +100 K instances. | 100,000 | documents | Medium | nan | nan | nan | https://link.springer.com/content/pdf/10.1007/s10579-022-09577-5.pdf | Arab | No | GitHub | Free | nan | Yes | reading comprehension | nan | nan | journal | nan | nan | nan | nan | Emad A. Alghamdi |
papluca/language-identification | [] | https://huggingface.co/datasets/papluca/language-identification | https://huggingface.co/datasets/papluca/language-identification | unknown | 2,021 | multilingual | ar-CLS: (Arabic (Classic)) | other | text | crawling | The Language Identification dataset is a collection of 90k samples consisting of text passages and corresponding language label. | 90,000 | sentences | Low | The Hugging Face Course Community Event | Multilingual Amazon Reviews Corpus, XNLI, and STSb Multi MT | nan | nan | Arab-Latn | No | HuggingFace | Free | nan | Yes | language identification | The Hugging Face Course Community Event | nan | nan | nan | nan | nan | nan | Khalid N. Elmadani |
Arabic News articles from Aljazeera.net | [] | nan | https://www.kaggle.com/datasets/arhouati/arabic-news-articles-from-aljazeeranet | unknown | 2,020 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | news articles | text | crawling | Natural Language Process, or NLP, is one of the most studied machine learning field. Much progress has been made in recent years which has allowed this field to move into large scale use in several domains. Nowadays, NLP is widely used in social networks, in search engines, in translation tools, in chatbot assistants, and many others cases … However, the progress and results differ from language to another. The majority of machine learning models treat in priority English and abandon the other languages, in particular Arabic. The main reason for this is the lack of datasets. Hence all the interest of the current datasets gathering +5000 news articles in Arabic. | 5,870 | documents | nan | nan | nan | nan | nan | Arab | No | kaggle | Free | nan | No | summarization , embedding ,classification model to identify article domain,sentiment classification, co-reference identification model, named-entity recognition | nan | nan | nan | nan | ABDELKADER RHOUATI | nan | nan | Kamel GAANOUN |
Quranic Arabic Corpus | [] | nan | https://corpus.quran.com/download/ | custom | 2,017 | ar | ar-CLS: (Arabic (Classic)) | other | text | other | morphology of quranic corpus | 128,218 | tokens | Low | nan | nan | nan | nan | Latn | Yes | other | Free | nan | No | morphological analysis | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
Shakkelha | [] | https://huggingface.co/datasets/arbml/shakkelha | https://github.com/AliOsm/shakkelha | MIT License | 2,019 | ar | ar-CLS: (Arabic (Classic)) | books | text | other | Arabic text diacritization extension dataset that is should be used for training only. This dataset is an extension of the dataset provided here: https://github.com/AliOsm/arabic-text-diacritization, and both of them were derived from the same source, which is Tashkeela dataset. | 533,000 | sentences | Low | Jordan University of Science and Technology (JUST) | Tashkeela | Neural Arabic Text Diacritization: State of the Art Results and a Novel Approach for Machine Translation | https://aclanthology.org/D19-5229/ | Arab | No | GitHub | Free | nan | No | Text diacritization | EMNLP-IJCNLP | 14.0 | workshop | Workshop on Asian Translation (WAT) | Ali Fadel, Ibraheem Tuffaha, Bara’ Al-Jawarneh, and Mahmoud Al-Ayyoub | Jordan University of Science and Technology (JUST) | In this work, we present several deep learning models for the automatic diacritization of Arabic text. Our models are built using two main approaches, viz. Feed-Forward Neural Network (FFNN) and Recurrent Neural Network (RNN), with several enhancements such as 100-hot encoding, embeddings, Conditional Random Field (CRF) and Block-Normalized Gradient (BNG). The models are tested on the only freely available benchmark dataset and the results show that our models are either better or on par with other models, which require language-dependent post-processing steps, unlike ours. Moreover, we show that diacritics in Arabic can be used to enhance the models of NLP tasks such as Machine Translation (MT) by proposing the Translation over Diacritization (ToD) approach. | Ali Hamdi Ali Fadel |
ashaar | [] | https://huggingface.co/datasets/MagedSaeed/ashaar | https://huggingface.co/datasets/MagedSaeed/ashaar | unknown | 2,022 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | other | text | crawling | ashaar: a dataset for Arabic poetry | 254,000 | documents | Low | nan | nan | nan | nan | Arab | No | HuggingFace | Free | nan | No | meter classification, poetry era classification, poetry theme classification | nan | nan | nan | nan | Maged S. Alshaibani, Zaid Alyafeai | King Fahud University of Petroleum and Minerals | nan | Maged S. Alshaibani |
Arabic Treebank - Weblog | [] | nan | https://catalog.ldc.upenn.edu/LDC2016T02 | LDC User Agreement | 2,016 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | web pages | text | crawling | This release contains 243,117 source tokens before clitics were split, and 308,996 tree tokens after clitics were separated for treebank annotation. The source material is weblogs collected by LDC from various sources. | 243,117 | tokens | Low | University of Pennsylvania | nan | Consistent and Flexible Integration of Morphological Annotation in the Arabic Treebank | https://catalog.ldc.upenn.edu/docs/LDC2016T02/KulickBiesMaamouri-LREC2010.pdf | Arab-Latn | No | LDC | With-Fee | 4,500.00 $ | No | part of speech taggin | LREC | 24.0 | conference | International Conference on Language Resources and Evaluation (LREC 2010). | Mohamed Maamouri, Ann Bies, Seth Kulick, Sondos Krouna, Dalila Tabassi, Michael Ciul | Linguistic Data Consortium - University of Pennsylvania | Complications arise for standoff annotation when the annotation is not on the source text itself, but on a more abstract representation.
This is particularly the case in a language such as Arabic with morphological and orthographic challenges, and we discuss various
aspects of these issues in the context of the Arabic Treebank. The Standard Arabic Morphological Analyzer (SAMA) is closely
integrated into the annotation workflow, as the basis for the abstraction between the explicit source text and the more abstract token
representation. However, this integration with SAMA gives rise to various problems for the annotation workflow and for maintaining
the link between the Treebank and SAMA. In this paper we discuss how we have overcome these problems with consistent and more
precise categorization of all of the tokens for their relationship with SAMA. We also discuss how we have improved the creation of
several distinct alternative forms of the tokens used in the syntactic trees. As a result, the Treebank provides a resource relating the
different forms of the same underlying token with varying degrees of vocalization, in terms of how they relate (1) to each other, (2) to
the syntactic structure, and (3) to the morphological analyzer. | Mustafa Ghaleb |
Arabic-Dataset-for-CAPT | [] | https://huggingface.co/datasets/arbml/CAPT | https://github.com/bhalima/Arabic-Dataset-for-CAPT | unknown | 2,015 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | transcribed audio | spoken | other | The dataset includes both “correct” and “wrong” non-artificial pronunciations. The pronunciations are from nine pupils aged from 5 to 8 years; each of them uttered a set of 16 sequences (words or group of words). The chosen words included some difficulties to learners such as the long vowels and the words written with more than one connected component. | 143 | sentences | Low | University Badji Mokhtar | nan | A statistical-based decision for arabic pronunciation assessment | https://link.springer.com/article/10.1007/s10772-014-9248-2 | Arab | No | GitHub | Free | nan | No | speech recognition | IJST | nan | journal | International Journal of Speech Technology | Khaled Necibi, Halima Bahi | Khaled Necibi & Halima Bahi | The aim of a computer assisted language learning (CALL) system is to improve the language skills of learners. Such systems often include, grammar and vocabulary components, while the pronunciation learning seems to be the hardest step in language learning process. Little attention has been paid to this aspect among the required ones in CALL systems. In pronunciation learning context, the learner would like to know if its pronunciation is good or bad. In the case where the pronunciation is bad, it will be suitable if some advices are given to him. The goal of this work is an early detection of pupils with reading difficulties and in the issue of decision whether their pronunciation is good or not is our particular interest. For this purpose, we consider the answer to this question as a classification problem and we use a statistical approach to make a decision; this approach allows us to pursue the investigation concerning the pronunciation of every phoneme in the word or in the sentence. | Zaid Alyafeai |
BAVED | [] | https://huggingface.co/datasets/arbml/BAVED | https://github.com/40uf411/Basic-Arabic-Vocal-Emotions-Dataset | unknown | 2,019 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | transcribed audio | spoken | manual curation | Basic Arabic Vocal Emotions Dataset (BAVED) is a datasetthat contains an arabic words spelled in diffrent levels of emotions recorded in an audio/wav format | 1,935 | tokens | Medium | nan | nan | nan | nan | Arab | No | GitHub | Free | nan | No | speech emotion recognition | nan | nan | nan | nan | nan | nan | nan | Emad A. Alghamdi |
Arabic Natural Audio Dataset | [] | nan | https://data.mendeley.com/datasets/xm232yxf7t/1 | CC BY 4.0 | 2,018 | ar | mixed | transcribed audio | spoken | other | This is the first Arabic Natural Audio Dataset (ANAD) developed to recognize 3 discrete emotions: Happy,angry, and surprised. | 1,384 | hours | Medium | Lebanese International University | nan | Emotion recognition in Arabic speech | https://link.springer.com/content/pdf/10.1007/s10470-018-1142-4.pdf | Arab | No | other | Free | nan | No | speech emotion recognition | nan | 30.0 | journal | Analog Integrated Circuits and Signal Processing | Samira Klaylat, Ziad Osman, Lama Hamandi, Rached Zantout | nan | Automatic emotion recognition from speech signals without linguistic cues has been an important emerging research area. Integrating emotions in human–computer interaction is of great importance to effectively simulate real life scenarios. Research has been focusing on recognizing emotions from acted speech while little work was done on natural real life
utterances. English, French, German and Chinese corpora were used for that purpose while no natural Arabic corpus was found to date. In this paper, emotion recognition in Arabic spoken data is studied for the first time. A realistic speech corpus from Arabic TV shows is collected. The videos are labeled by their perceived emotions; namely happy, angry or
surprised. Prosodic features are extracted and thirty-five classification methods are applied. Results are analyzed in this paper and conclusions and future recommendations are identified. | Abdelrahman Rezk |
Arabic Sentiment Twitter Corpus | [] | nan | https://www.kaggle.com/datasets/mksaad/arabic-sentiment-twitter-corpus | custom | 2,019 | ar | mixed | social media | text | crawling and annotation(other) | A Sentiment Analysis dataset. No extra information is provided regarding the dialects nor the collection methodology | 58,000 | sentences | Medium | nan | nan | nan | nan | Arab | No | kaggle | Free | nan | Yes | sentiment analysis | nan | nan | nan | nan | Motaz Saad | nan | nan | Maged S. Alshaibani |
Arabic - Egyptian comparable Wikipedia corpus | [] | https://huggingface.co/datasets/arbml/Comparable_Wikipedia | https://www.kaggle.com/datasets/mksaad/arb-egy-cmp-corpus | CC BY-SA 4.0 | 2,017 | ar | mixed | wikipedia | text | crawling | The dataset is composed of a set of text documents in both Arabic (Modern Standard) and Egyptian dialect aligned at document level. comparable documents share the same document ID. | nan | documents | Low | nan | nan | WikiDocsAligner: An Off-the-Shelf Wikipedia Documents Alignment Tool | https://ieeexplore.ieee.org/document/8038320 | Arab | No | kaggle | Free | nan | No | dialect identification, text generation, language modeling | PICICT | nan | conference | Palestinian International Conference on Information and Communication Technology | Motaz Saad, Basem O. Alijla | Islamic University of Gaza | Wikipedia encyclopedia is an attractive source for comparable corpora in many languages. Most researchers develop their own script to perform document alignment task, which requires efforts and time. In this paper, we present WikiDocsAligner, an off-the-shelf Wikipedia Articles alignment handy tool. The implementation of WikiDocsAligner does not require the researchers to import/export of interlanguage links databases. The user just need to download Wikipedia dumps (interlanguage links and articles), then provide them to the tool, which performs the alignment. This software can be used easily to align Wikipedia documents in any language pair. Finally, we use WikiDocsAligner to align comparable documents from Arabic Wikipedia and Egyptian Wikipedia. So we shed the light on Wikipedia as a source of Arabic dialects language resources. The produced resources is interesting and useful as the demand on Arabic/dialects language resources increased in the last decade. | Zaid Alyafeai |
Lebanon Uprising Arabic Tweets | [] | nan | https://www.kaggle.com/datasets/abedkhooli/lebanon-uprising-october-2019-tweets | unknown | 2,019 | multilingual | mixed | social media | text | crawling | This is a collection of tweets related to the Arabic hashtag (#لبنان_ينتفض) on Lebanon uprising in October 2019. | 100,000 | sentences | High | nan | nan | nan | nan | Arab-Latn | No | kaggle | Free | nan | No | sentiment analysis | nan | nan | nan | nan | ABED KHOOLI | nan | nan | Mustafa Ghaleb |
wikimedia/wit_base | [] | https://huggingface.co/datasets/wikimedia/wit_base | https://github.com/google-research-datasets/wit | CC BY-SA 4.0 | 2,021 | multilingual | ar-MSA: (Arabic (Modern Standard Arabic)) | wikipedia | text | crawling | Wikimedia's version of the Wikipedia-based Image Text (WIT) Dataset, a large multimodal multilingual dataset. | 6,477,255 | sentences | Low | Google | nan | WIT: Wikipedia-based Image Text Dataset for Multimodal Multilingual Machine Learning | https://arxiv.org/pdf/2103.01913.pdf | Arab-Latn | No | HuggingFace | Free | nan | No | image captioning, text retrieval | SIGIR '21 | 40.0 | conference | Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval | Krishna Srinivasan, Karthik Raman, Jiecao Chen, Michael Bendersky, Marc Najork | Google | The milestone improvements brought about by deep representation learning and pre-training techniques have led to large performance gains across downstream NLP, IR and Vision tasks. Multimodal modeling techniques aim to leverage large high-quality visio-linguistic datasets for learning complementary information (across image and text modalities). In this paper, we introduce the Wikipedia-based Image Text (WIT) Dataset (this https URL) to better facilitate multimodal, multilingual learning. WIT is composed of a curated set of 37.6 million entity rich image-text examples with 11.5 million unique images across 108 Wikipedia languages. Its size enables WIT to be used as a pretraining dataset for multimodal models, as we show when applied to downstream tasks such as image-text retrieval. WIT has four main and unique advantages. First, WIT is the largest multimodal dataset by the number of image-text examples by 3x (at the time of writing). Second, WIT is massively multilingual (first of its kind) with coverage over 100+ languages (each of which has at least 12K examples) and provides cross-lingual texts for many images. Third, WIT represents a more diverse set of concepts and real world entities relative to what previous datasets cover. Lastly, WIT provides a very challenging real-world test set, as we empirically illustrate using an image-text retrieval task as an example. | Khalid N. Elmadani |
Arabic News Tweets | [] | https://huggingface.co/datasets/arbml/Arabic_News_Tweets | https://data.mendeley.com/datasets/9dxgbgx86k/3 | CC BY 4.0 | 2,021 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | social media | text | crawling and annotation(other) | This dataset is a relatively great size collection of Arabic news tweets that were collected from an official and verified users in Twitter. All news that is collected from the most popular and official users in Saudi Arabia belongs to Saudi Arabia news. All data that is gathered was retrieved using specific time period and collected all news in that time. To the best of our knowledge, this dataset is the first Arabic news data collection that does not specify by keywords and belongs to Saudi Arabia. This news dataset can be valuable for diverse tasks in NLP, such as text classification and automated verification system. The dataset has been categorized into 5 different news classes which are general news, regions news, sport news, economic news, and quality life news. In this data article, 89,179 original tweets have presented and fully labeled into related categories. | 89,179 | sentences | nan | Umm Al-Qura University, University of Technology Sydney | nan | nan | nan | Arab | No | Mendeley Data | Free | nan | No | topic classification | nan | nan | nan | nan | Karali Sami, Thanoon Mohammed, Lin Chin-Teng | nan | nan | Kamel GAANOUN |
OSCAR-2201 | [] | https://huggingface.co/datasets/oscar-corpus/OSCAR-2201 | https://huggingface.co/datasets/oscar-corpus/OSCAR-2201 | CC0 | 2,022 | multilingual | mixed | web pages | text | crawling | OSCAR or Open Super-large Crawled Aggregated coRpus is a huge multilingual corpus obtained by language classification and filtering of the Common Crawl corpus using the ungoliant architecture. Data is distributed by language in both original and deduplicated form. | 8,718,929 | documents | Low | nan | Common Crawl | Towards a Cleaner Document-Oriented Multilingual Crawled Corpus | https://arxiv.org/pdf/2201.06642.pdf | Arab | No | HuggingFace | Upon-Request | nan | No | text generation, language modeling | arXiv | nan | preprint | nan | Julien Abadji, Pedro Ortiz Suarez, Laurent Romary, Benoıt Sagot
| nan | The need for raw large raw corpora has dramatically increased in recent years with the introduction of transfer learning and
semi-supervised learning methods to Natural Language Processing. And while there have been some recent attempts to
manually curate the amount of data necessary to train large language models, the main way to obtain this data is still through
automatic web crawling. In this paper we take the existing multilingual web corpus OSCAR and its pipeline Ungoliant that
extracts and classifies data from Common Crawl at the line level, and propose a set of improvements and automatic annotations
in order to produce a new document-oriented version of OSCAR that could prove more suitable to pre-train large generative
language models as well as hopefully other applications in Natural Language Processing and Digital Humanities.
| Zaid Alyafeai |
DataSet for Arabic Classification | [] | https://huggingface.co/datasets/arbml/DataSet_Arabic_Classification | https://data.mendeley.com/datasets/v524p5dhpj/2 | CC BY 4.0 | 2,018 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | news articles | text | crawling | DataSet for Arabic text classification. The dataset, as mentioned by the author's description has been collected semi-automatically. | 111,700 | documents | Low | Universite Sultan Moulay Slimane de Beni-Mellal, Universite Chouaib Doukkali Faculte des Sciences | nan | nan | nan | Arab | No | Mendeley Data | Free | nan | No | topic classification | nan | nan | nan | nan | mohamed BINIZ | nan | nan | Maged S. Alshaibani |
CORMAP: Corpus for Moroccan Arabic Processing | [] | nan | https://lindat.mff.cuni.cz/repository/xmlui/handle/11372/LRT-3551 | LGPL-3.0 | 2,017 | ar | ar-MA: (Arabic (Morocco)) | other | text | other | This resource is a corpus containing 34k Moroccan Colloquial Arabic sentences collected from different sources. The sentences are written in Arabic letters. This resource can be useful in some NLP applications such as Language Identification. | 34,000 | sentences | nan | LINDAT/CLARIAH-CZ | nan | nan | nan | Arab | No | other | Free | nan | No | language identification | nan | nan | nan | nan | tachicart ridouane ,bouzoubaa, karim | nan | nan | Kamel GAANOUN |
araData | [
{
"Name": "GLF",
"Dialect": "ar-GLF: (Arabic (Gulf))",
"Volume": "2,007",
"Unit": "nan"
},
{
"Name": "General",
"Dialect": "mixed",
"Volume": "2,003",
"Unit": "nan"
},
{
"Name": "EGY",
"Dialect": "ar-EG: (Arabic (Egypt))",
"Volume": "2,002",
"Unit": "nan"
},
{
"Name": "LEV",
"Dialect": "ar-LEV: (Arabic(Levant))",
"Volume": "2,001",
"Unit": "nan"
},
{
"Name": "MGH",
"Dialect": "ar-NOR: (Arabic (North Africa))",
"Volume": "2,001",
"Unit": "nan"
},
{
"Name": "Tunisian",
"Dialect": "ar-TN: (Arabic (Tunisia))",
"Volume": "2,001",
"Unit": "nan"
},
{
"Name": "IRQ",
"Dialect": "ar-IQ: (Arabic (Iraq))",
"Volume": "216",
"Unit": "nan"
}
] | https://huggingface.co/datasets/arbml/arData | https://github.com/malek-hedhli/araData | unknown | 2,022 | ar | mixed | other | text | other | araData is a clean and balanced dataset containing sentences for 7 Arabic dialects represented as follows: 2007 GLF, 2003 general, 2002 EGY, 2001 LEV, 2001 MGH, 2001 Tunisian, and 216 IRQ. This data is a collection of different resources prepared by HEDHLI Malek for a project to identify Arabic dialects. | 12,236 | sentences | Low | nan | nan | nan | nan | Arab | No | GitHub | Free | nan | No | dialect identification | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
BOLT Arabic Discussion Forums | [] | nan | https://catalog.ldc.upenn.edu/LDC2018T10 | LDC User Agreement | 2,018 | ar | mixed | other | text | crawling | BOLT Arabic Discussion Forums was developed by the Linguistic Data Consortium (LDC) and consists of 813,080 discussion forum threads in Egyptian Arabic harvested from the Internet using a combination of manual and automatic processes. | 813,080 | documents | Medium | University of Pennsylvania | nan | BOLT Arabic Discussion Forums | https://catalog.ldc.upenn.edu/LDC2018T10 | Arab | No | other | With-Fee | nan | No | machine translation | nan | nan | nan | nan | Jennifer Tracey, Haejoong Lee, Stephanie Strassel, Safa Ismael | nan | nan | Abdelrahman Rezk |
khalidalt/tydiqa-goldp | [] | https://huggingface.co/datasets/khalidalt/tydiqa-goldp | https://huggingface.co/datasets/khalidalt/tydiqa-goldp | unknown | 2,020 | multilingual | ar-CLS: (Arabic (Classic)) | other | text | crawling | TyDi QA is a question answering dataset covering 11 typologically diverse languages with 204K question-answer pairs. | 204,000 | documents | Low | Google | nan | TyDi QA: A Benchmark for Information-Seeking Question Answering in Typologically Diverse Languages | https://aclanthology.org/2020.tacl-1.30.pdf | Arab-Latn | No | HuggingFace | Free | nan | Yes | question answering | TACL | 178.0 | conference | Transactions of the Association for Computational Linguistics | Jonathan H. Clark, Eunsol Choi, Michael Collins, Dan Garrette, Tom Kwiatkowski, Vitaly Nikolaev, Jennimaria Palomaki | Google | Confidently making progress on multilingual modeling requires challenging, trustworthy evaluations. We present TyDi QA—a question answering dataset covering 11 typologically diverse languages with 204K question-answer pairs. The languages of TyDi QA are diverse with regard to their typology—the set of linguistic features each language expresses—such that we expect models performing well on this set to generalize across a large number of the world’s languages. We present a quantitative analysis of the data quality and example-level qualitative linguistic analyses of observed language phenomena that would not be found in English-only corpora. To provide a realistic information-seeking task and avoid priming effects, questions are written by people who want to know the answer, but don’t know the answer yet, and the data is collected directly in each language without the use of translation. | Khalid N. Elmadani |
ANAD: Arabic Natural Audio Dataset | [] | https://huggingface.co/datasets/arbml/ANAD | https://data.mendeley.com/datasets/xm232yxf7t/1 | CC BY 4.0 | 2,018 | ar | mixed | transcribed audio | spoken | manual curation | Eight videos of live calls between an anchor and a human outside the studio were downloaded from online Arabic talk shows. Each video was then divided into turns: callers and receivers. To label each video, 18 listeners were asked to listen to each video and select whether they perceive a happy, angry or surprised emotion. Silence, laughs and noisy chunks were removed. Every chunk was then automatically divided into 1 sec speech units forming our final corpus composed of 1384 records. | 1,384 | hours | High | nan | nan | nan | nan | Arab | No | Mendeley Data | Free | nan | No | emotion recognition | nan | 14.0 | nan | nan | Samira klaylat, ziad Osman, Rached Zantout, Lama Hamandi | nan | nan | Mustafa Ghaleb |
ArabCeleb | [] | nan | https://github.com/CeLuigi/ArabCeleb | CC BY 4.0 | 2,021 | ar | mixed | transcribed audio | spoken | crawling | ArabCeleb is an audio dataset collected in the wild that specifically focuses on arabic language. The proposed dataset contains 1930 utterances from 100 celebrities taken from video on YouTube.com. The dataset might be used for several speaker recognition tasks: identification, verification, gender recognition as well as multimodal recognition tasks thus integrating audio and video tracks. | 1,930 | sentences | Low | nan | nan | ArabCeleb: Speaker Recognition in Arabic | nan | Arab | No | GitHub | Free | nan | Yes | speech recognition | AIxIA | nan | conference | International Conference of the Italian Association for Artificial Intelligence | Bianco, Simone and Celona, Luigi and Khalifa, Intissar and Napoletano, Paolo and Petrovsky, Alexey and Piccoli, Flavio and Schettini, Raimondo and Shanin, Ivan | nan | nan | Zaid Alyafeai |
Arabic Stop words | [] | https://huggingface.co/datasets/arbml/arabic_stop_words | https://github.com/mohataher/arabic-stop-words | MIT License | 2,016 | ar | mixed | other | text | other | Largest list of Arabic stop words on Github. | 750 | tokens | Low | nan | nan | nan | nan | Arab | Yes | GitHub | Free | nan | No | sentiment analysis | nan | nan | nan | nan | Mohamed Taher Alrefaie, Tarek BAZINE | nan | nan | Abdelrahman Rezk |
Arabic Wiki data Dump 2018 | [] | nan | https://www.kaggle.com/datasets/abedkhooli/arabic-wiki-data-dump-2018 | unknown | 2,018 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | wikipedia | text | crawling | Arabic is a rich and major world language. Recent advances in computational linguistics and AI can be applied to Arabic but not in the generic way most languages are treated. This dataset (Arabic articles from Wikipedia) will be used to train Word2Vec and compare performance with publicly available pre-trained model from FastText (Facebook) in a generic way. A related model is now available: https://www.kaggle.com/abedkhooli/arabic-ulmfit-model | nan | documents | nan | nan | nan | nan | nan | Arab | No | kaggle | Free | nan | No | text generation, language modeling | nan | nan | nan | nan | ABED KHOOLI | nan | nan | Kamel GAANOUN |
Ajdir Corpora | [] | nan | http://aracorpus.e3rab.com/argistestsrv.nmsu.edu/AraCorpus/ | unknown | 2,010 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | news articles | text | crawling | This is a raw text from Arabic daily newspapers collected over a year between 2004 and 2005. Each file is compiled as cleaned raw text from documents that are separated by two blank lines. | 28 | documents | Low | nan | nan | nan | nan | Arab | No | other | Free | nan | No | text generation, language modeling | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
Sentiment Lexicons for 81 Languages | [] | https://huggingface.co/datasets/senti_lex | https://www.kaggle.com/datasets/rtatman/sentiment-lexicons-for-81-languages | unknown | 2,017 | multilingual | mixed | other | text | crawling | Sentiment analysis, the task of automatically detecting whether a piece of text is positive or negative, generally relies on a hand-curated list of words with positive sentiment (good, great, awesome) and negative sentiment (bad, gross, awful). This dataset contains both positive and negative sentiment lexicons for 81 languages. | 2,794 | tokens | Low | nan | nan | nan | nan | Arab | Yes | GitLab | Free | nan | No | sentiment classification
| nan | nan | nan | nan | RACHAEL TATMAN | nan | nan | Abdelrahman Rezk |
West Point Arabic Speech | [] | nan | https://catalog.ldc.upenn.edu/LDC2002S02 | LDC User Agreement | 2,002 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | transcribed audio | spoken | manual curation | The corpus consists of 8,516 speech files, totaling 1.7 gigabytes or 11.42 hours of speech data. Each speech file represents one person reciting one prompt from one of four prompt scripts. The utterances were recorded using a Shure SM10A microphone and a RANE Model MS1 pre-amplifier. The files were recorded as 16-bit PCM low-byte-first ("little-endian") raw audio files, with a sampling rate of 22.05 KHz. They were then converted to NIST sphere format. Approximately 7,200 of the recordings are from native informants and 1200 files are from non-native informants. The following tables show the breakdown of corpus content in terms of male, female, native and non-native speakers. | 11.42 | hours | Low | nan | nan | nan | nan | Arab | No | LDC | Free | 1,000.00 $ | No | speech recognition | nan | nan | nan | nan | Stephen A. LaRocca, Rajaa Chouairi | Department of Foreign languages at the United States Military Academy at West Point and the Center For Technology Enhanced Language Learning (CTELL) | nan | Mustafa Ghaleb |
Sudanese Dialect tweets about telecommunication companies | [] | https://huggingface.co/datasets/arbml/Sudanese_Dialect_Tweet_Tele | https://docs.google.com/spreadsheets/d/13fIV8oHss-QRBKN-2h5LYF1i_1O9qH1R/edit?usp=sharing&ouid=101796411348671465142&rtpof=true&sd=true | unknown | 2,018 | ar | ar-SD: (Arabic (Sudan)) | social media | text | crawling and annotation(other) | Sentiment Analysis dataset written in Sudanese Arabic Dialect | 4,712 | sentences | High | University of Khartoum | nan | Sentiment Analysis for Arabic Dialect Using Supervised Learning | https://ieeexplore.ieee.org/document/8515862 | Arab | No | Gdrive | Free | nan | No | sentiment analysis | ICCCEEE | 8.0 | conference | International Conference on Computer, Control, Electrical, and Electronics Engineering 2018 | Rua Ismail, Mawada Omer, Mawada Tabir, Noor Mahadi, Izzeldein Amin | University of Khartoum | Sentiment analysis is a set of procedures used to extract subjective opinions from the text. Generally, there are two techniques for sentiment analysis, machine learning method, and lexicon-based method. This work focuses on extracting and analyzing Twitter data written in Sudanese Arabic dialect to observe opinionated patterns regarding the quality of telecommunication services operating in Sudan. One of the significant limitations in the field of text classification is the exclusive focus on the English language. There is a need to bridge this gap by developing efficient methods and tools for sentiment analysis in the Arabic language. Moreover, reliable corpus and lexicons are needed. For this study, four classifiers were trained on a dataset consist of 4712 tweets. Namely Naïve Bayes, SVM, Multinomial Logistic Regression and K-Nearest Neighbor to conduct a comparative analysis on the performance of the classifiers. These algorithms when ran against the tweets dataset the results revealed that SVM gives the highest F1-score (72.0) while the best accuracy was achieved by KNN (k=2) and it equals to 92.0. | Khalid N. Elmadani |
SERAG: Semantic Entity Retrieval from Arabic knowledge Graphs | [] | https://huggingface.co/datasets/arbml/ArabicDEv2 | https://zenodo.org/record/4560653#.YprApXZBxD8 | CC BY 4.0 | 2,021 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | other | text | human translation | This is the dataset used in: "S. Esmeir, SERAG: Semantic Entity Retrieval from Arabic knowledge Graphs, In: Proceedings of the Sixth Arabic Natural Language Processing Workshop (WANLP 2021) at EACL 2021". The dataset is a translation of a subset (139/467) of the queries in DBpedia Entity v2 to Modern Standard Arabic. We used the “stopped” version of DBpedia Entity v2 (queries-v2_stopped.txt). Please use the query ID to link the translated queries to their English counterparts, and to the relevance judgment files provided with DBpedia Entity v2 (qrels-v2.txt). DBpedia’s interlingual mapping file (interlanguage_links_ar.ttl.bz2) can be used to map entities from English to Arabic and vice-versa. | 139 | sentences | nan | nan | DBpedia Entity v2 | SERAG: Semantic Entity Retrieval from Arabic Knowledge Graphs | https://aclanthology.org/2021.wanlp-1.24/ | Arab | No | zenodo | Free | nan | No | semantic entity retrieval | ACL | nan | workshop | The Sixth Arabic Natural Language Processing Workshop (WANLP 2021), | Esmeir Saher | Bloomberg L.P | Knowledge graphs (KGs) are widely used to store and access information about entities and their relationships. Given a query, the task of entity retrieval from a KG aims at presenting a ranked list of entities relevant to the query. Lately, an increasing number of models for entity retrieval have shown a significant improvement over traditional methods. These models, however, were developed for English KGs. In this work, we build on onesuchsystem, named KEWER, to propose SERAG (Semantic Entity Retrieval from Arabic knowledge Graphs). Like KEWER, SERAG uses random walks to generate entity embeddings. DBpedia-Entity v2 is considered the standard test collection for entity retrieval. We discuss the challenges of using it for non-English languages in general and Arabic in particular. We provide an Arabic version of this standard collection, and use it to evaluate SERAG. SERAG is shown to significantly outperform the popular BM25 model thanks to its multi-hop reasoning. | Kamel GAANOUN |
inaracorpus | [] | nan | https://sourceforge.net/projects/inaracorpus/ | unknown | 2,013 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | other | text | other | InAra corpus comprises 1024 documents; 80% of them contain passages borrowed from other documents to simulate plagiarism. For each suspicious document an XML file is associated ; it contains the length and the position of each plagiarism passage. | 1,024 | documents | Low | nan | nan | nan | nan | Arab | No | sourceforge | Free | nan | No | Intrinsic plagiarism detection | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
Names transliteration | [] | https://huggingface.co/datasets/arbml/names_transliteration | https://github.com/thomas-chauvet/names_transliteration | unknown | 2,020 | multilingual | ar-MSA: (Arabic (Modern Standard Arabic)) | other | text | other | containing names in arabic characters and associated names in latin characters (english) | 118,047 | tokens | Low | nan | ANETAC dataset, Google transliteration data, NETransliteration-COLING2018 | nan | nan | Arab | No | GitHub | Free | nan | No | machine translation | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
QAC: Qatari Arabic Corpus | [] | nan | http://www.isle.illinois.edu/dialect/QAC/index.html | CC BY 4.0 | 2,014 | ar | ar-QA: (Arabic (Qatar)) | transcribed audio | spoken | crawling | Speech was recorded from four Qatari television programs in 2009-2011: | 18.45 | hours | Low | Qatar University | nan | Development of a TV Broadcasts Speech Recognition System for Qatari Arabic | http://www.lrec-conf.org/proceedings/lrec2014/pdf/430_Paper.pdf | Arab | No | other | Free | nan | No | speech recognition | LREC | nan | conference | Language Resources and Evaluation Conference | Mohamed Elmahdy, Mark Hasegawa-Johnson, Eiman Mustafawi | Qatar University; University of Illinois ;Qatar University | A major problem with dialectal Arabic speech recognition is due to the sparsity of speech resources. In this paper, a transfer learning
framework is proposed to jointly use a large amount of Modern Standard Arabic (MSA) data and little amount of dialectal Arabic data to
improve acoustic and language modeling. The Qatari Arabic (QA) dialect has been chosen as a typical example for an under-resourced
Arabic dialect. A wide-band speech corpus has been collected and transcribed from several Qatari TV series and talk-show programs.
A large vocabulary speech recognition baseline system was built using the QA corpus. The proposed MSA-based transfer learning
technique was performed by applying orthographic normalization, phone mapping, data pooling, acoustic model adaptation, and system
combination. The proposed approach can achieve more than 28% relative reduction in WER.
| Zaid Alyafeai |
NEMLAR Written Corpus | [] | nan | https://catalogue.elra.info/en-us/repository/browse/ELRA-W0042/ | Non Commercial Use - ELRA END USER | 2,006 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | transcribed audio | spoken | other | The NEMLAR Written Corpus consists of about 500,000 words of Arabic text from 13 different categories, aiming to achieve a well-balanced corpus that offers a representation of the variety in syntactic, semantic and pragmatic features of modern Arabic language | 500,000 | tokens | Low | nan | nan | nan | nan | Arab | No | ELRA | With-Fee | 300.00€ | No | speech recognition | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
Tunisian Arabic Corpus | [] | nan | http://www.tunisiya.org/ | unknown | 2,010 | ar | ar-TN: (Arabic (Tunisia)) | web pages | text | crawling | There are currently 2,874 texts in the corpus, comprising 1,088,614 words. The top categories currently included are displayed below. As you can see, the internet sources are currently dominant | 2,874 | documents | Low | nan | nan | Tunisian Arabic Corpus: Creating a written corpus of an " unwritten " language | https://www.academia.edu/28966672/Tunisian_Arabic_Corpus_Creating_a_written_corpus_of_an_unwritten_language | Arab | No | other | Free | nan | No | morphological analysis | nan | nan | preprint | nan | Karen McNeil | nan | nan | Zaid Alyafeai |
ArVox | [] | nan | https://www.kaggle.com/datasets/corpora4research/arpod-corpus-based-on-arabic-podcasts | unknown | 2,019 | multilingual | mixed | transcribed audio | spoken | crawling | designed for Multilingual and Arabic Dialect Identification | nan | hours | Low | nan | nan | A Language Identification System Based on Voxforge Speech Corpus | https://link.springer.com/chapter/10.1007/978-3-030-14118-9_53 | Arab | No | kaggle | Free | nan | No | dialect identification from speech | AMLTA | nan | conference | International Conference on Advanced Machine Learning Technologies and Applications | Khaled Lounnas, Mourad Abbas, Hocine Teffahi, Mohamed Lichouri | University of Sciences and Technology Houari Boumediene; CRSTDLA;University of Sciences and Technology Houari Boumediene; University of Sciences and Technology Houari Boumediene | In this work, we address the problem of identifying languages based on Voxforge speech corpus. We downloaded corpora for three languages: English, German and Persian from Voxforge. In addition, we recorded two additional corpora, the first one for Modern Standard Arabic (MSA) and the other one for Kabyl, one of the Algerian Berber dialects. To tackle this task, we used three classifiers, namely: k-Nearest Neighbors (kNN), Support Vector Machines (SVM) and Extra Trees Classifier. We obtained an average precision of 87.45% for binary classification compared to 44% for the multi-class one. | Zaid Alyafeai |
ArPod | [
{
"Name": "KSA",
"Dialect": "ar-SA: (Arabic (Saudi Arabia))",
"Volume": "2.33",
"Unit": "hours"
},
{
"Name": "MSA",
"Dialect": "ar-MSA: (Arabic (Modern Standard Arabic))",
"Volume": "0.83",
"Unit": "hours"
},
{
"Name": "SYR",
"Dialect": "ar-SY: (Arabic (Syria))",
"Volume": "0.83",
"Unit": "hours"
},
{
"Name": "EGY",
"Dialect": "ar-EG: (Arabic (Egypt))",
"Volume": "1.5",
"Unit": "hours"
},
{
"Name": "LEB",
"Dialect": "ar-LB: (Arabic (Lebanon))",
"Volume": "1.5",
"Unit": "hours"
},
{
"Name": "ENG",
"Dialect": "mixed",
"Volume": "0.83",
"Unit": "hours"
}
] | nan | https://www.kaggle.com/datasets/corpora4research/arpod-corpus-based-on-arabic-podcasts | unknown | 2,019 | multilingual | mixed | other | spoken | other | The corpus has a duration of 8.1 hours sampled at 16 kHz and coded on 16 bits. The considered languages are: MSA, English; whereas the dialects are: Syrian, Saudi Arabic, Egyptian, and Lebanese. 70% of this dataset had been used for training, and the remaining for the test phase. | 7.82 | hours | Low | nan | nan | Building a Speech Corpus based on Arabic Podcasts for Language and Dialect Identification | https://aclanthology.org/W19-7408.pdf | Arab | No | kaggle | Free | nan | No | dialect classification from speech | ICNLSP | nan | conference | International Conference on Natural Language and Speech Processing | Khaled Lounnas, Mourad Abbas, Mohamed Lichouri
| USTHB University, Algeria; CRSTDLA;CRSTDLA | In this paper, we present ArPod, a new Arabic speech
corpus made of Arabic audio podcasts. We built this
dataset, mainly for both speech-based multi-lingual and
multi-dialectal identification tasks. It includes two languages: Modern Standard Arabic (MSA) and English,
and four Arabic dialects: Saudi, Egyptian, Lebanese
and Syrian. A set of supervised classifiers have been
used: Support Vector Machines (SVM), Multi Layer
Perceptron (MLP), K-Nearest Neighbors (KNN), Extratrees and Convolutional Neural Networks (CNN),
using acoustic and spectral features. For both tasks,
SVM yielded encouraging results and outperformed the
other classifiers. Language Identification, Dialect
Identification, CNN, Acoustic features, spectral features, SVM, Arabic Podcast
| Zaid Alyafeai |
PAN18 Author Profiling | [] | nan | https://zenodo.org/record/3746006#.YptKU3ZBxD8 | unknown | 2,018 | multilingual | mixed | social media | text | crawling and annotation(other) | We provide you with a training data set that consists of Twitter users labeled with gender. For each author, a total of 100 tweets and 10 images are provided. Authors are grouped by the language of their tweets: English, Arabic and Spanish. | 250,000 | sentences | nan | Bauhaus-Universität Weimar | nan | Overview of the 6th Author Profiling Task at PAN 2018: Multimodal Gender Identification in Twitter | https://pan.webis.de/downloads/publications/papers/rangel_2018.pdf | Arab-Latn | No | zenodo | Upon-Request | nan | Yes | author profiling | CLEF | 132.0 | conference | Conference and Labs of the Evaluation Forum | Rangel, Francisco, Rosso, Paolo, Potthast, Martin, & Stein, Benno. | nan | nan | Nouamane Tazi |
PAN17 Author Profiling | [] | nan | https://zenodo.org/record/3745980#.YqTxWnZBxD9 | unknown | 2,017 | multilingual | mixed | social media | text | crawling and annotation(other) | We provide you with a training data set that consists of Twitter tweets in English, Spanish, Portuguese and Arabic, labeled with gender and language variety. | 4,000 | sentences | nan | Bauhaus-Universität Weimar | nan | Overview of PAN’17 Author Identification, Author Profiling, and Author Obfuscation | https://riunet.upv.es/bitstream/handle/10251/102943/PAN-2017-author.pdf | Arab-Latn | No | zenodo | Upon-Request | nan | Yes | author profiling | CLEF | 85.0 | conference | nan | nan | nan | nan | Nouamane Tazi |
The Arabic Speech Corpus for Isolated Words | [] | nan | https://www.cs.stir.ac.uk/~lss/arabic/ | unknown | 2,014 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | transcribed audio | spoken | manual curation | The Arabic speech corpus for isolated words contains 9992 utterances of 20 words spoken by 50 native male Arabic speakers. It has been recorded with a 44100 Hz sampling rate and 16-bit resolution. This corpus is free for noncommercial uses in the raw format (.wav files) and other formats e.g. (MFCCs) are available under request. | 9,992 | tokens | Low | University of Stirling | nan | On Improving the Classification Capability of Reservoir Computing for Arabic Speech Recognition | https://www.cs.stir.ac.uk/~lss/recentpapers/icann2014AlalshekmubarakSmith.pdf | Arab | No | other | Free | nan | No | speech recognition | nan | nan | preprint | nan | Abdulrahman Alalshekmubarak, Leslie S. Smith | University of Stirling | Designing noise-resilient systems is a major challenge in the
field of automated speech recognition (ASR). These systems are crucial
for real-world applications where high levels of noise tend to be present.
We introduce a noise robust system based on Echo State Networks and
Extreme Kernel machines which we call ESNEKM. To evaluate the performance of the proposed system, we used our recently released public
Arabic speech dataset and the well-known spoken Arabic digits (SAD)
dataset. Different feature extraction methods considered in this study
include mel-frequency cepstral coefficients (MFCCs), perceptual linear
prediction (PLP) and RASTA- perceptual linear prediction. These extracted features were fed to the ESNEKM and the result compared with
a baseline hidden Markov model (HMM), so that nine models were compared in total. ESNEKM models outperformed HMM models under all
the feature extraction methods, noise levels, and noise types. The best
performance was obtained by the model that combined RASTA-PLP
with ESNEKM. | Zaid Alyafeai |
Merged Arabic Corpus of Isolated Words | [] | nan | https://www.kaggle.com/datasets/mohamedanwarvic/merged-arabic-corpus-of-isolated-words | ODbL-1.0 | 2,019 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | transcribed audio | spoken | other | This dataset is a voice-recorded dataset of 50 Native-Arabic speakers saying 20 words about 10 times. It has been recorded with a 44100 Hz sampling rate and 16-bit resolution. This dataset can be used for tasks like Speaker Recognition, Speaker Verification, Voice biometrics, | 9,992 | tokens | Low | nan | The Arabic Speech Corpus for Isolated Words | nan | nan | Arab | No | kaggle | Free | nan | No | speaker recognition, speech recognition | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
HC Corpora | [] | nan | https://web.archive.org/web/20161021044006/http://corpora.heliohost.org/ | unknown | 2,016 | multilingual | mixed | web pages | text | crawling | The corpora are collected from publicly available sources by a web crawler. The crawler checks for language, so as to mainly get texts consisting of the desired language | nan | documents | Low | nan | nan | nan | nan | Arab | No | other | Free | nan | No | text generation, language modeling | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
Arabic DEv2 | [] | https://huggingface.co/datasets/arbml/ArabicDEv2 | https://zenodo.org/record/4560653#.YqSGWXZBxD9 | CC BY 4.0 | 2,021 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | other | text | human translation | The dataset is a translation of a subset (139/467) of the queries in DBpedia Entity v2 (https://github.com/iai-group/DBpedia-Entity) to Modern Standard Arabic. We used the “stopped” version of DBpedia Entity v2 (queries-v2_stopped.txt). | 139 | sentences | Low | Bloomberg L.P. | DEv2 | SERAG: Semantic Entity Retrieval from Arabic knowledge Graphs | https://aclanthology.org/2021.wanlp-1.24.pdf | Arab | No | zenodo | Free | nan | No | entity retrieval from knowledge graphs | WANLP | nan | workshop | Workshop on Arabic Natural Language Processing | Saher Esmeir | Bloomberg L.P. | Knowledge graphs (KGs) are widely used to
store and access information about entities and
their relationships. Given a query, the task of
entity retrieval from a KG aims at presenting
a ranked list of entities relevant to the query.
Lately, an increasing number of models for entity retrieval have shown a significant improvement over traditional methods. These models,
however, were developed for English KGs. In
this work, we build on one such system, named
KEWER, to propose SERAG (Semantic Entity Retrieval from Arabic knowledge Graphs).
Like KEWER, SERAG uses random walks to
generate entity embeddings. DBpedia-Entity
v2 is considered the standard test collection
for entity retrieval. We discuss the challenges
of using it for non-English languages in general and Arabic in particular. We provide an
Arabic version of this standard collection, and
use it to evaluate SERAG. SERAG is shown
to significantly outperform the popular BM25
model thanks to its multi-hop reasoning.
| Zaid Alyafeai |
Arabic Broadcast News Transcripts | [] | nan | https://catalog.ldc.upenn.edu/LDC2006T20 | LDC User Agreement | 2,001 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | transcribed audio | spoken | other | Arabic Broadcast News Transcripts was developed by the Linguistic Data Consortium (LDC) and consists of 10 hours of transcribed speech from Voice of America satellite radio news broadcasts in Arabic recorded by LDC between June 2000 and January 2001. | 10 | hours | Low | LDC | nan | nan | nan | Arab | No | LDC | With-Fee | 400.00 $ | No | speech recognition | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
ACE 2004 Multilingual Training Corpus | [] | nan | https://catalog.ldc.upenn.edu/LDC2005T09 | LDC User Agreement | 2,004 | multilingual | ar-MSA: (Arabic (Modern Standard Arabic)) | other | text | other | ACE 2004 Multilingual Training Corpus was developed by the Linguistic Data Consortium (LDC) and contains the various genre text in English (158,000 words), Chinese (307,000 characters, 154,000 words), and Arabic (151,000 words) annotated for entities and relations. | 689 | documents | Low | LDC | nan | nan | nan | Arab | No | LDC | With-Fee | 3,000.00 $ | No | named entity recognition | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
Iraqi Arabic Conversational Telephone Speech | [] | nan | https://catalog.ldc.upenn.edu/LDC2006S45 | LDC User Agreement | 2,006 | ar | ar-IQ: (Arabic (Iraq)) | transcribed audio | spoken | other | Iraqi Arabic Conversational Telephone Speech was developed by Appen Pty Ltd, Sydney, Australia and contains roughly 3000 mins of speech from Iraqi Arabic speakers taking part in spontaneous telephone conversations in Colloquial Iraqi Arabic. | 50 | hours | Low | LDC | nan | nan | nan | Arab | No | LDC | Free | nan | No | speech recognition | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
CTAB: Corpus of Tunisian Arabizi | [] | https://huggingface.co/datasets/arbml/CTAB | https://zenodo.org/record/4781769#.YqSPY3ZBxD9 | CC BY 4.0 | 2,021 | ar | ar-TN: (Arabic (Tunisia)) | social media | text | crawling | This dataset has been created between 2017 and 2021 to provide a textual resource that can be used to study the behaviors of Tunisian people in writing Tunisian Arabic (ISO 693-3: aeb) in Latin Script. This corpus is constituted from messages written using Tunisian Arabic Chat Alphabet or Arabizi and is developed to solve the matter of the lack of NLP databases about the use of the Latin Script for transcribing Tunisian Arabic. | 5,702 | sentences | Low | University of Sfax | nan | nan | nan | Arab | No | zenodo | Free | nan | No | dialect identification | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
Levantine Arabic Conversational Telephone Speech, Transcripts | [] | nan | https://catalog.ldc.upenn.edu/LDC2007T01 | LDC User Agreement | 2,007 | ar | ar-LEV: (Arabic(Levant)) | transcribed audio | spoken | other | This database contains 982 Levantine Arabic speakers taking part in spontaneous telephone conversations in Colloquial Levantine Arabic. A total of 985 conversation sides are provided (there are three speakers who each appear in two disctinct conversations). The average duration per side is between 5 and 6 minutes. | 985 | sentences | nan | Appen Pty Ltd | nan | nan | nan | Arab | No | LDC | With-Fee | 200.00 $ | No | speech recognition | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
Fisher Levantine Arabic Conversational Telephone Speech | [] | nan | https://catalog.ldc.upenn.edu/LDC2007S02 | LDC User Agreement | 2,006 | ar | ar-LEV: (Arabic(Levant)) | transcribed audio | spoken | other | Levantine Arabic QT Training Data Set 5, Speech was developed by the Linguistic Data Consortium (LDC) and contains 1,660 calls totalling approximately 250 hours of telephone conversation in Levantine Arabic. | 250 | hours | Low | LDC | nan | nan | nan | Arab | No | LDC | With-Fee | 4,000.00 $ | No | speech recognition | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
ARPC: A Corpus for Paraphrase Identification in Arabic Text | [] | nan | https://ieee-dataport.org/documents/arpc-corpus-paraphrase-identification-arabic-text#files | unknown | 2,019 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | other | text | other | ArPC is an Arabic paraphrase identification corpus. It consists of 1331 sentence pairs along with their binary score that indicates weather the pairs are paraphrase or not. The corpus has been manually annotated by three Arabic native speakers. | 1,331 | sentences | Low | nan | nan | nan | nan | Arab | No | other | With-Fee | nan | Yes | paraphrase identification | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
CSLU: 22 Languages Corpus | [] | nan | https://catalog.ldc.upenn.edu/LDC2005S26 | LDC User Agreement | 2,002 | multilingual | mixed | transcribed audio | spoken | other | Produced by Center for Spoken Language Understanding and distributed by the Linguistic Data Consortium, the 22 Languages corpus consists of telephone speech from 21 languages: Eastern Arabic, Cantonese, Czech, Farsi, German, Hindi, Hungarian, Japanese, Korean, Malay, Mandarin, Italian, Polish, Portuguese, Russian, Spanish, Swedish, Swahili, Tamil, Vietnamese, and English. The corpus contains fixed vocabulary utterances (e.g. days of the week) as well as fluent continuous speech. Each of the 50,191 utterances is verified by a native speaker to determine if the caller followed instructions when answering the prompts. For this release, approximately 19,758 utterances have corresponding orthographic transcriptions in all the above languages except Eastern Arabic, Farsi, Korean, Russian, Italian. | 84 | hours | Low | LDC | nan | nan | nan | Arab | No | LDC | With-Fee | 150.00 $ | No | speech recognition | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
TDT4 Multilingual Broadcast News Speech Corpus | [] | nan | https://catalog.ldc.upenn.edu/LDC2005S11 | LDC User Agreement | 2,005 | multilingual | ar-MSA: (Arabic (Modern Standard Arabic)) | transcribed audio | spoken | other | The TDT4 corpus contains news data collected daily from 20 news sources (13 broadcast, seven newswire) in three languages (American English, Mandarin Chinese, and Modern Standard Arabic), over a period of four months (October 2000 through January 2001). Here's a breakdown of the broadcast data included in this release with number of files and time length by source: | 88.3 | hours | Low | LDC | nan | nan | nan | Arab | No | LDC | With-Fee | 19,000.00 $ | No | speech recognition | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
Arab-ESL | [] | https://huggingface.co/datasets/arbml/emoji_sentiment_lexicon | https://github.com/ShathaHakami/Arabic-Emoji-Sentiment-Lexicon-Version-1.0/blob/main/Arabic_Emoji_Sentiment_Lexicon_Version_1.0.csv | Non Commercial Use - ELRA END USER | 2,021 | ar | mixed | social media | text | crawling and annotation(other) | Emoji (the popular digital pictograms) are sometimes seen as a new kind of artificial and universally usable and consistent writing code. In spite of their assumed universality, there is some evidence that the sense of an emoji, specifically in regard to sentiment, may change from language to language and culture to culture. This paper investigates whether contextual emoji sentiment analysis is consistent across Arabic and European languages. To conduct this investigation, we, first, created the Arabic emoji sentiment lexicon (Arab-ESL). Then, we exploited an existing European emoji sentiment lexicon to compare the sentiment conveyed in each of the two families of language and culture (Arabic and European). The results show that the pairwise correlation between the two lexicons is consistent for emoji that represent, for instance, hearts, facial expressions, and body language. However, for a subset of emoji (those that represent objects, nature, symbols, and some human activities), there are large differences in the sentiment conveyed. More interestingly, an extremely high level of inconsistency has been shown with food emoji. | 1,034 | tokens | Low | Jazan University / University of Birmingham | nan | Arabic Emoji Sentiment Lexicon (Arab-ESL): A Comparison between Arabic and European Emoji Sentiment Lexicons | https://aclanthology.org/2021.wanlp-1.7/ | Arab | Yes | GitHub | Free | nan | No | sentiment analysis | WANLP | 1.0 | workshop | Arabic Natural Language Processing Workshop | Shatha Ali A. Hakami, Robert Hendley, Phillip Smith | Jazan University / University of Birmingham | Emoji (the popular digital pictograms) are sometimes seen as a new kind of artificial and universally usable and consistent writing code. In spite of their assumed universality, there is some evidence that the sense of an emoji, specifically in regard to sentiment, may change from language to language and culture to culture. This paper investigates whether contextual emoji sentiment analysis is consistent across Arabic and European languages. To conduct this investigation, we, first, created the Arabic emoji sentiment lexicon (Arab-ESL). Then, we exploited an existing European emoji sentiment lexicon to compare the sentiment conveyed in each of the two families of language and culture (Arabic and European). The results show that the pairwise correlation between the two lexicons is consistent for emoji that represent, for instance, hearts, facial expressions, and body language. However, for a subset of emoji (those that represent objects, nature, symbols, and some human activities), there are large differences in the sentiment conveyed. More interestingly, an extremely high level of inconsistency has been shown with food emoji. | Shatha Ali A. Hakami |
Levantine Arabic QT Training Data Set 5, Speech | [] | nan | https://catalog.ldc.upenn.edu/LDC2006S29 | LDC User Agreement | 2,006 | ar | ar-LEV: (Arabic(Levant)) | transcribed audio | spoken | other | Levantine Arabic QT Training Data Set 5, Speech was developed by the Linguistic Data Consortium (LDC) and contains 1,660 calls totalling approximately 250 hours of telephone conversation in Levantine Arabic. These calls were collected between 2003 and 2005. | 250 | hours | Low | LDC | nan | nan | nan | Arab | No | LDC | With-Fee | 4,000.00 $ | No | speech recognition | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
Arab-Andalusian music corpus | [] | nan | https://zenodo.org/record/1291776#.YqTFeHZBxD9 | CC BY-NC 4.0 | 2,018 | ar | mixed | transcribed audio | spoken | other | The following files are available for 164 concert recordings (overall playable time more than 125 hours): | 125 | hours | Low | nan | nan | nan | nan | Arab | No | zenodo | Free | nan | No | speech classification | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
Arabic Infectious Disease Ontology | [] | nan | http://www.research.lancs.ac.uk/portal/en/datasets/arabic-infectious-disease-ontology(39dbef60-ae9b-4405-99c8-35a41e95a3e0).html | unknown | 2,020 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | web pages | text | crawling | This file contains an Arabic Infectious Disease Ontology to include Non-Standard Terminology. The Arabic Infectious Disease Ontology is written in the Arabic language, and is the first ontology in Arabic specialising in the infectious disease domain. It contains 11 classes, 21 object properties, 11 datatype properties, and 215 individual concepts. | 215 | sentences | Low | Lancaster University | nan | Developing an Arabic Infectious Disease Ontology to Include Non-Standard Terminology | https://eprints.lancs.ac.uk/id/eprint/142307/1/LREC_2020_Paper_Developing_an_Arabic_Infectious_Ontology_.pdf | Arab | Yes | other | Free | nan | No | infectious disease ontology | LREC | nan | conference | Language Resources and Evaluation Conference | Lama Alsudias, Paul Rayson | Lancaster University; King Saud University | Building ontologies is a crucial part of the semantic web endeavour. In recent years, research interest has grown rapidly in supporting
languages such as Arabic in NLP in general but there has been very little research on medical ontologies for Arabic. We present a new
Arabic ontology in the infectious disease domain to support various important applications including the monitoring of infectious disease
spread via social media. This ontology meaningfully integrates the scientific vocabularies of infectious diseases with their informal
equivalents. We use ontology learning strategies with manual checking to build the ontology. We applied three statistical methods for
term extraction from selected Arabic infectious diseases articles: TF-IDF, C-value, and YAKE. We also conducted a study, by consulting
around 100 individuals, to discover the informal terms related to infectious diseases in Arabic. In future work, we will automatically
extract the relations for infectious disease concepts but for now these are manually created. We report two complementary experiments
to evaluate the ontology. First, a quantitative evaluation of the term extraction results and an additional qualitative evaluation by a
domain expert. | Zaid Alyafeai |
AlRiyadh-Newspaper-Covid-Dataset | [] | https://huggingface.co/datasets/arbml/AlRiyadh_Newspaper_Covid | https://github.com/alioh/AlRiyadh-Newspaper-Covid-Dataset | CC BY 3.0 | 2,021 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | news articles | text | crawling | It is a dataset of Arabic newspapers articles addressing COVID-19 related events. The data origin is Alriyadh newspaper. It contains all news articles until 1 February 2021 | 24,084 | documents | Low | nan | nan | nan | nan | Arab | No | GitHub | Free | nan | No | topic analysis (covid19) | nan | nan | nan | nan | nan | nan | nan | Emad A Alghamdi |
ARL Arabic Dependency Treebank | [] | nan | https://catalog.ldc.upenn.edu/LDC2016T18 | LDC User Agreement | 2,016 | ar | ar-MSA: (Arabic (Modern Standard Arabic)) | news articles | text | other | The source data in this release consists of Arabic newswire and broadcast programming collected by LDC from various news and broadcast providers. | nan | tokens | Low | LDC | Arabic Treebank | nan | nan | Arab | Yes | LDC | With-Fee | 2,000.00 $ | No | part of speech tagging | nan | nan | nan | nan | nan | nan | nan | Zaid Alyafeai |
emoji-sentiment-dataset | [] | nan | https://github.com/snakers4/emoji-sentiment-dataset/tree/master#dataset | custom | 2,019 | multilingual | ar-MSA: (Arabic (Modern Standard Arabic)) | social media | text | crawling | Following the success of DeepMoji and TorchMoji (1, 2), we would like to leverage Twitter as an open source of self-annotated data to create a balanced multi-language "in-the-wild" sentiment dataset to test the quality of various NLP models and/or word/sub-word tokenization techniques. | 287,578 | tokens | Medium | nan | nan | nan | nan | Arab-Latn | Yes | GitHub | Free | nan | No | sentiment analysis | nan | nan | nan | nan | nan | nan | nan | Emad A Alghamdi |
Transliteration | [] | https://huggingface.co/datasets/arbml/google_transliteration | https://github.com/google/transliteration | Apache-2.0 | 2,016 | multilingual | ar-MSA: (Arabic (Modern Standard Arabic)) | wikipedia | text | crawling | Arabic-English transliteration dataset mined from Wikipedia. | 15,898 | tokens | Low | Google | nan | Sequence-to-sequence neural network models for transliteration | https://arxiv.org/pdf/1610.09565.pdf | Arab | No | GitHub | Free | nan | Yes | transliteration | arXiv | nan | preprint | nan | Mihaela Rosca, Thomas Breuel | Google | Transliteration is a key component of machine
translation systems and software internationalization. This paper demonstrates that neural
sequence-to-sequence models obtain state of
the art or close to state of the art results on existing datasets. In an effort to make machine
transliteration accessible, we open source a
new Arabic to English transliteration dataset
and our trained models. | Zaid Alyafeai |
NETransliteration | [] | https://huggingface.co/datasets/arbml/NETransliteration | https://github.com/steveash/NETransliteration-COLING2018 | MIT License | 2,018 | multilingual | ar-MSA: (Arabic (Modern Standard Arabic)) | wikipedia | text | crawling | data files mined from wikidata | 145,186 | sentences | Low | Amazon | nan | Design Challenges in Named Entity Transliteration | https://aclanthology.org/C18-1053.pdf | Arab | No | GitHub | Free | nan | No | transliteration, named entity recognition | COLING | nan | preprint | International Conference on Computational Linguistics | Yuval Merhav, Stephen Ash | Amazon Alexa AI; Amazon AWS AI | We analyze some of the fundamental design challenges that impact the development of a multilingual state-of-the-art named entity transliteration system, including curating bilingual named entity datasets and evaluation of multiple transliteration methods. We empirically evaluate the transliteration task using the traditional weighted finite state transducer (WFST) approach against two neural approaches: the encoder-decoder recurrent neural network method and the recent, non-sequential Transformer method. In order to improve availability of bilingual named entity transliteration datasets, we release personal name bilingual dictionaries mined from Wikidata for English to Russian, Hebrew, Arabic, and Japanese Katakana. Our code and dictionaries are publicly available. | Zaid Alyafeai |