SinaLab
Research Group (ontology + NLP) at Birzeit University, Palestine
https://sina.birzeit.edu
20/05/2026
๐ข Announcement: Shared task at ArabicNLP at EMNLP 2026:
ู
ุณุงุจูุฉ ุจุญุซูุฉ - ูุฏุฑุฉ ุงูุฐูุงุก ุงูุงุตุทูุงุนู ุนูู ุงูุชุนู
ูู
ู
ุน ุงูุจูุงูุงุช ุบูุฑ ุงูู
ุฑุฆูุฉ ู
ุณุจูุง
๐๐ฟ๐ฎ๐๐ฒ๐ป๐ฟ๐ฒ: ๐ ๐๐ถ๐ฒ๐ฟ๐ฎ๐ฟ๐ฐ๐ต๐ถ๐ฐ๐ฎ๐น ๐๐ฒ๐ณ๐ถ๐ป๐ถ๐๐ถ๐ผ๐ป-๐๐๐ถ๐ฑ๐ฒ๐ฑ ๐๐ฟ๐ฎ๐ฏ๐ถ๐ฐ ๐๐ฒ๐ป๐ฟ๐ฒ ๐๐น๐ฎ๐๐๐ถ๐ณ๐ถ๐ฐ๐ฎ๐๐ถ๐ผ๐ป
Participate: https://wp.lancs.ac.uk/aragenre/
A challenging task designed to test how NLP systems generalise to the unseen:
โขโ โ Few-shot training and development data
โขโ โ Cross-lingual genre definitions guiding Arabic classification
โขโ โ A large hidden test set of naturally occurring Arabic texts
โขโ โ Unseen genre-definition combinations
While AraGenre focuses on Arabic, the task addresses a wider challenge in NLP: building methods that can generalise from limited data and support languages and varieties that remain underrepresented in current NLP research.
Organisers:
Mo El-Haj
Saad Ezzini
Mustafa Jarrar
Shadi Abudalfa
#๐๐ฟ๐ฎ๐๐ฒ๐ป๐ฟ๐ฒ
19/05/2026
ุฅูุฌุงุฒ ุฃูุงุฏูู
ู ู
ุชู
ูุฒ
ููุฎุฑ ุจุทูุจุชูุง
ูููุฆ ุงูุทุงูุจุฉ ุชูู
ุงุก ุญู
ูุฏุฉ ูู ุจุฑูุงู
ุฌ ู
ุงุฌุณุชูุฑ ููุฏุณุฉ ุงูุจุฑู
ุฌูุงุช ูุจูู ูุดุฑ ูุฑูุชูุง ุงูุจุญุซูุฉ ุจุนููุงู:
"AraREQ: A Dataset and End-to-End System for Conflict Detection and Resolution in Software Requirements"
ูุฐูู ูู ู
ุคุชู
ุฑ
International Conference on Language Resources and Evaluation (LREC 2026)
ุฃุญุฏ ุฃุจุฑุฒ ุงูู
ุคุชู
ุฑุงุช ุงูุฏูููุฉ ุงูู
ุชุฎุตุตุฉ ูู ู
ุฌุงู ู
ูุงุฑุฏ ุงููุบุฉ ูุชูููู
ูุง.
ูุฌุงุกุช ูุฐู ุงููุฑูุฉ ุงูุจุญุซูุฉ ู
ุณุชูุฉ ู
ู ุฑุณุงูุฉ ุงูู
ุงุฌุณุชูุฑ ุจุฅุดุฑุงู ุงูุฏูุชูุฑ ู
ุตุทูู ุฌุฑุงุฑ ุจุนููุงู:
"An End-to-End System for Detecting and Resolving Conflicts in Arabic Software Requirements Using LLMs"
ูุชูุงููุช ุงูุฏุฑุงุณุฉ ุจูุงุก ู
ุฌู
ูุนุฉ ุจูุงูุงุช ุนุฑุจูุฉ ู
ุชุฎุตุตุฉ ูููุดู ุนู ุงูุชุนุงุฑุถุงุช ูู ู
ุชุทูุจุงุช ุงูุจุฑู
ุฌูุงุชุ ุญูุซ ุชู
ุชูุณูู
ุฃูุซุฑ ู
ู 26,545 ุฒูุฌูุง ู
ู ุงูู
ุชุทูุจุงุช ุจุดูู ูุฏูู ูููุดู ุนู ุงูุชุนุงุฑุถุงุช ูุชุตููููุง.
ูู
ุง ุชู
ุชุทููุฑ ูุธุงู
ู
ุชูุงู
ู ูุนุชู
ุฏ ุนูู ูู
ุงุฐุฌ ุงููุบุฉ ุงููุจูุฑุฉ (LLMs) ูููุดู ุนู ุงูุชุนุงุฑุถุงุช ูุงูุชุฑุงุญ ุญููู ููุงุ ู
ุน ุงุฎุชุจุงุฑ ุนุฏุฉ ูู
ุงุฐุฌ ู
ุซู:ChatGPTุ ูLLaMAุ ูFanarุ ูDeepSeekุ ููุฏ ุญูู ุงููุธุงู
ุฏูุฉ ุนุงููุฉ ุจูุบุช 94% F1-score.
ููุง ููุชุตุฑ ุฃุซุฑ ูุฐุง ุงูุนู
ู ุนูู ู
ุฌุงู ููุฏุณุฉ ุงูุจุฑู
ุฌูุงุช ูุญุณุจุ ุจู ูู
ุชุฏ ูุฏุนู
ุฃุจุญุงุซ ู
ุนุงูุฌุฉ ุงููุบุฉ ุงูุนุฑุจูุฉ (Arabic NLP) ุงูู
ุชุนููุฉ ุจูุดู ุงูุชุนุงุฑุถุงุช ูุญููุง ูู ู
ุฌุงูุงุช ู
ุชุนุฏุฏุฉ.
ูู
ูู ููู
ูุชู
ูู ุงูุงุทูุงุน ุนูู ุงููุฑูุฉ ุงูุจุญุซูุฉ ุนูู ุงูุฑุงุจุท ูู ุงูุชุนููู ุงูุงูู
ูู ุงูุฏุนู
ูุงูุชูุฏูุฑ ููุทุงูุจุฉ ุชูู
ุงุก ููู
ุดุฑููุง ุฏูุชูุฑ ุฌุฑุงุฑุ ูู
ุฒูุฏุงู ู
ู ุงูุชูุฏู
ูุงููุฌุงุญ ูุงูุชู
ูุฒ ุงูุฃูุงุฏูู
ู
08/11/2025
Join us at to learn ๐๐จ๐ฐ ๐ญ๐จ ๐ฎ๐ฌ๐ ๐จ๐ง๐ญ๐จ๐ฅ๐จ๐ ๐ข๐๐ฌ ๐ข๐ง ๐๐๐ ๐ฉ๐ซ๐จ๐ฆ๐ฉ๐ญ๐ฌ?
We are excited to release our ๐๐จ๐ฃ๐จ๐จ๐๐๐ง๐ญ๐จ๐ฅ๐จ๐ ๐ฒ for ๐๐๐ฆ๐๐ ๐๐ง๐ญ๐ข๐ญ๐ฒ ๐๐๐๐จ๐ ๐ง๐ข๐ญ๐ข๐จ๐ง and ๐๐๐ฅ๐๐ญ๐ข๐จ๐ง ๐๐ฑ๐ญ๐ซ๐๐๐ญ๐ข๐จ๐ง at at .
ูุณุนุฏูุง ุงูุฅุนูุงู ุนู "ุฃูุทูููุฌูุง ูุฌูุฏ"ุ ูุชูุญูุฏ ุทุฑู ุงุณุชุฎุฑุงุฌ ุฃุณู
ุงุก ุงูุฃุนูุงู
ูุงูุนูุงูุงุช ุจูููุงุ ููุฐูู ุงุณุชุฎุฏุงู
ุงูุงูุทูููุฌูุง ุฃุซูุงุก ุงุณุชุฎุฏุงู
ุงููู
ุงุฐุฌ ุงููุบููุฉ ุงูุถุฎู
ุฉ.
- Paper: https://lnkd.in/dyVydKvN
- Download: https://lnkd.in/dYDka4sw
- YouTube: https://lnkd.in/dsN2mTrs
- Slides: https://lnkd.in/dQJm55S4
๐๐ก๐จ๐ซ๐ญ ๐ฌ๐ฎ๐ฆ๐ฆ๐๐ซ๐ฒ:
Information Extraction tasks such as Named Entity Recognition and Relation Extraction often rely on diverse tagsets and annotation guidelines, creating challenges for model generalization, cross-dataset evaluation, and interoperability. To overcome these issues, WojoodOntology provides a unified semantic framework that aligns heterogeneous tagsets and guidelines.
It supports two ontology-based mapping strategies: rule-based unidirectional alignment and ontology-based prompting, where ontology concepts are embedded directly into large language model prompts to enable more effective bidirectional mappings. Experiments demonstrate that ontology-based prompting improves out-of-domain mapping accuracy by 15% compared to rule-based methods. Additionally, WojoodOntology is aligned with Schema.org and Wikidata, ensuring compatibility with knowledge graphs and facilitating wider industry adoption.
#๐๐ก๐๐ซ๐๐๐๐๐ฌ๐ค
Alaa Aljabari Nagham Hamad Mohammed Khalilia Mustafa Jarrar Birzeit University
07/11/2025
Article: ๐ซ๐๐ฅ๐๐ญ๐ข๐จ๐ง ๐๐ฑ๐ญ๐ซ๐๐๐ญ๐ข๐จ๐ง ๐๐จ๐ซ๐ฉ๐ฎ๐ฌ. ~550K tokens, 40 relations.
ูุดุฑ ูุฑูุฉ ูู ู
ุคุชู
ุฑ ู
ุฑู
ูู ููุทุงูุจุฉ ุฃูุงุก ุงูุฌุนุจุฑู - ุจุฑูุงู
ุฌ ุงูุฏูุชูุฑุงุฉ ุจุนูู
ุงูุญุงุณูุจ. ุงููุฑูุฉ ุชูุฏู
ู
ููุฌูุฉ ู
ุฏููุฉ ุฌุฏูุฏุฉ ูุงุณุชุฎุฑุงุฌ ุงูุนูุงูุงุช ุจูู ุฃุณู
ุงุก ุงูุฃุนูุงู
ูู ุงููุตูุต ุงูุนุฑุจูุฉ.
Article:https://www.jarrar.info/publications/JKJ25.pdf
Download: https://sina.birzeit.edu/relations/
Demo: https://sina.birzeit.edu/relations/
Slides:https://www.jarrar.info/Talks/wojood_relations_paper.pdf
๐ฌ๐ฎ๐ฆ๐ฆ๐๐ซ๐ฒ: We introduce WojoodRelations, the largest Arabic relation extraction corpus, containing 33K sentences and 15K annotated relation triples across 40 distinct relation types. We also propose new methods for supervised and LLM-based RE, with our models achieving up to 92.9% F1, setting strong baselines for future research.
SinaLab
ุงูุฏุฑุงุณุงุช ุงูุนููุง ูุงูุฃุจุญุงุซ ูู ุฌุงู
ุนุฉ ุจูุฑุฒูุช
ูููุฉ ุงูููุฏุณุฉ -ุฌุงู
ุนุฉ ุจูุฑุฒูุช
Mustafa Jarrar
Hamad Bin Khalifa University
Birzeit University | ุฌุงู
ุนุฉ ุจูุฑุฒูุช
PhD in Computer Science / ุฏูุชูุฑุงุฉ ุนูู
ุญุงุณูุจ
05/11/2025
Our 7 articles at in China, (Join us!):
ุณุจุนุฉ ุฃูุฑุงู ุณุชุนุฑุถ ูู ุงูู
ุคุชู
ุฑ ุงูุฏููู ูุฃุณุงููุจ ุงูู
ุนุงูุฌุฉ ุงููุบููุฉ ุงูุทุจูุนูุฉ ูู ุงูุตูู
1 ๐๐จ๐ฃ๐จ๐จ๐๐๐๐ฅ๐๐ญ๐ข๐จ๐ง๐ฌ: ๐๐ซ๐๐๐ข๐ ๐๐๐ฅ๐๐ญ๐ข๐จ๐ง ๐๐ฑ๐ญ๐ซ๐๐๐ญ๐ข๐จ๐ง ๐๐จ๐ซ๐ฉ๐ฎ๐ฌ ๐๐ง๐ ๐๐จ๐๐๐ฅ๐ข๐ง๐ .https://www.jarrar.info/publications/JKJ25.pdf
at
2 ๐๐๐ญ๐ข๐ฏ๐ ๐๐๐๐ซ๐ง๐ข๐ง๐ ๐๐จ๐ซ ๐๐ฎ๐ฅ๐ญ๐ข๐๐ข๐๐ฅ๐๐๐ญ๐๐ฅ ๐๐ซ๐๐๐ข๐ ๐๐๐ ๐๐๐ ๐ ๐ข๐ง๐ .https://www.jarrar.info/publications/AKJ25.pdf
at
3 ๐๐๐๐ซ๐ฅ: ๐ ๐๐ฎ๐ฅ๐ญ๐ข๐ฆ๐จ๐๐๐ฅ ๐๐ฎ๐ฅ๐ญ๐ฎ๐ซ๐๐ฅ๐ฅ๐ฒ-๐๐ฐ๐๐ซ๐ ๐๐ซ๐๐๐ข๐ ๐๐ง๐ฌ๐ญ๐ซ๐ฎ๐๐ญ๐ข๐จ๐ง ๐๐๐ญ๐๐ฌ๐๐ญ.https://www.jarrar.info/publications/AME25.pdf
at
4 ๐๐จ๐ฃ๐จ๐จ๐๐๐ง๐ญ๐จ๐ฅ๐จ๐ ๐ฒ: ๐๐ง๐ญ๐จ๐ฅ๐จ๐ ๐ฒ-๐๐ซ๐ข๐ฏ๐๐ง ๐๐๐ ๐๐ซ๐จ๐ฆ๐ฉ๐ญ๐ข๐ง๐ ๐๐จ๐ซ ๐๐ง๐ข๐๐ข๐๐ ๐๐ง๐๐จ๐ซ๐ฆ๐๐ญ๐ข๐จ๐ง ๐๐ฑ๐ญ๐ซ๐๐๐ญ๐ข๐จ๐ง ๐๐๐ฌ๐ค๐ฌ.https://www.jarrar.info/publications/JHKJ25.pdf
at
5 ๐๐ฆ๐๐ ๐๐๐ฏ๐๐ฅ ๐๐๐๐: ๐๐ก๐ ๐
๐ข๐ซ๐ฌ๐ญ ๐๐ซ๐๐๐ข๐ ๐๐ฆ๐๐ ๐ ๐๐๐ฉ๐ญ๐ข๐จ๐ง๐ข๐ง๐ ๐๐ก๐๐ซ๐๐ ๐๐๐ฌ๐คhttps://www.jarrar.info/publications/BAH25.pdf
at
6 ๐๐ก๐ ๐๐ซ๐๐๐๐ง๐๐ฏ๐๐ฅ ๐๐ก๐๐ซ๐๐ ๐๐๐ฌ๐ค ๐จ๐ง ๐๐ซ๐๐๐ข๐ ๐๐ฎ๐ญ๐ก๐จ๐ซ๐ฌ๐ก๐ข๐ฉ ๐๐ญ๐ฒ๐ฅ๐๐๐ซ๐๐ง๐ฌ๐๐๐ซ ๐๐ง๐ ๐๐ ๐๐๐ง๐๐ซ๐๐ญ๐๐ ๐๐๐ฑ๐ญ ๐๐๐ญ๐๐๐ญ๐ข๐จ๐งhttps://www.jarrar.info/publications/AEA25.pdf
at
7 ๐๐๐๐ ๐๐๐๐: ๐๐ก๐ ๐
๐ข๐ซ๐ฌ๐ญ ๐๐ฎ๐ฅ๐ญ๐ข๐๐ข๐๐ฅ๐๐๐ญ๐๐ฅ ๐๐ซ๐๐๐ข๐ ๐๐ฉ๐๐๐๐ก ๐๐ซ๐จ๐๐๐ฌ๐ฌ๐ข๐ง๐ ๐๐ก๐๐ซ๐๐ ๐๐๐ฌ๐คhttps://www.jarrar.info/publications/TTS25.pdf
at
#๐๐ก๐๐ซ๐๐๐๐๐ฌ๐ค
SinaLab
Birzeit University | ุฌุงู
ุนุฉ ุจูุฑุฒูุช
31/10/2025
ูุฑูุฉ: ุงูุชุดููุง ุงู ู
ุฏููุฉ ู
ู 15K ููู
ุฉ ุชููู ูุชุฏุฑูุจ ูู
ูุฐุฌ ููู ููุฌุฉ.
Spotlight Paper: ๐๐ฐ๐๐ถ๐๐ฒ ๐๐ฒ๐ฎ๐ฟ๐ป๐ถ๐ป๐ด ๐ณ๐ผ๐ฟ ๐ ๐๐น๐๐ถ๐ฑ๐ถ๐ฎ๐น๐ฒ๐ฐ๐๐ฎ๐น ๐๐ฟ๐ฎ๐ฏ๐ถ๐ฐ ๐ฃ๐ข๐ฆ ๐ง๐ฎ๐ด๐ด๐ถ๐ป๐ด, accepted at .
๐ annotating 15K tokens is enough for high accuracy in POS tagging.
Paper: https://lnkd.in/dMtnSd8a
Slides: https://lnkd.in/diBzkfzZ
๐๐ก๐จ๐ซ๐ญ ๐ฌ๐ฎ๐ฆ๐ฆ๐๐ซ๐ฒ: We introduce an active learning approach for multidialectal (Palestinian, Syrian, Lebanese, Egyptian, Emirati) POS tagging that cuts annotation costs while maintaining high accuracy. We also found that ~15K annotated tokens is enough to achieve very high accuracy (97%). See the paper for the details.
#๐๐ก๐๐ซ๐๐๐๐๐ฌ๐ค
SinaLab
Birzeit University | ุฌุงู
ุนุฉ ุจูุฑุฒูุช
Hamad Bin Khalifa University
AraGenEval: Arabic Authorship Style Transfer and AI Generated Text Detection Shared Task The rapid expansion of user-generated content across social media, digital news platforms, and online communication has created a growing demand for sophisticated Natural Language Processing (NLP) techniques to analyze and manipulate writing styles. Unlike general text style analysis [1], which focu...
11/06/2025
ุฏุนูุฉ ููู
ุดุงุฑูุฉ -ุงูู
ุณุงุจูุฉ ุงูุจุญุซูุฉ ูุชูุตูู ุงูุตูุฑ
ูู ุงูู
ูู
ุฉ ุงูุฃููู ุณูุชูุญ ุตูุฑ ููุชู
ูุตููุง ูุฏููุงู.
ูู ุงูู
ูู
ุฉ ุงูุซุงููุฉุ ุณูุชูุญ ุตูุฑ ูุฃูุตุงููุง ูุชุฏุฑูุจ ูู
ุงุฐุฌ ูู
ูููุง ูุตู ุงูุตูุฑ ุขููุงู.
https://sina.birzeit.edu/image_eval2025/index.html
๐๐๐ง๐๐ฆ๐๐ง: 4,000 non-captioned images (subtask-1); and 3500 captioned images in Arabic (subtask-2).
๐ฆ๐๐ฏ๐๐ฎ๐๐ธ๐ญ: Image captioning datathon - participants will caption manually, and compete on the quantity and quality of captions.
๐ฆ๐๐ฏ๐๐ฎ๐๐ธ๐ฎ: Participants will be given training and test datasets (images with their captions) to develop models to automatically generate descriptions. Teams will compete on the quality of the generated captions that will be assessed using the standard metrics BLEU, ROUGE, and LLMs as a judge.
๐๐๐ ๐ฟ๐ผ๐๐๐
June 1, 2025 โ Data release and registration open
July 20, 2025 โ Registration deadline & test set release
July 25, 2025 โ Test set submission deadline
July 30, 2025 โ Final results announced
August 15, 2025 โ System description paper submission
September 5, 2025 โ Camera-ready deadline
๐ข๐ฅ๐๐๐ก๐๐ญ๐๐ฅ๐ฆ
Ahlam Bash*ti, Birzeit University.
Alaa aljabari, Birzeit University.
Mustafa Jarrar, Hamad Bin Khalifa University/Birzeit University
Fadi Zaraket, Arab Center for Research and Policy Studies/American University of Beirut
Bilal Shalash, Arab Center for Research and Policy Studies
George Mikros, Hamad Bin Khalifa University
Wajdi Zaghouani Ph.D, Northwestern University in Qatar
Ehsaneddin Asgari, Hamad Bin Khalifa University
Birzeit University, American University of Beirut, Hamad Bin Khalifa University, Arab Center for Research and Policy Studies ุงูู
ุฑูุฒ ุงูุนุฑุจู ููุฃุจุญุงุซ ูุฏุฑุงุณุฉ ุงูุณูุงุณุงุช Northwestern University
hashtag hashtag hashtag hashtag hashtag hashtag hashtag hashtag hashtag hashtag hashtag hashtag hashtag hashtag hashtag hashtag hashtag hashtag
hashtag hashtag #๐๐ก๐๐ซ๐๐๐๐๐ฌ๐ค hashtag hashtag hashtag hashtag hashtag hashtag hashtag hashtag
Image Captioning Shared Task 2025 The Arabic Image Captioning Shared Task - to develop and evaluate open sources image captions in Arabic.
Click here to claim your Sponsored Listing.
Category
Contact the business
Telephone
Website
Address
Ramallah