Our Day 2 of #ArabicNLP2026 starts with a fireside chat. Join us if you’re in Budapest.
@_ArabicNLP
"Which language is this?" sounds like the easiest question in NLP. For most of Africa's languages, it still has no reliable answer.
Today/Tuesday at #EMNLP2026, @SangyooonKwon and I will present AFROSCOPE, our framework for identifying African languages at scale.
The gap is real. Existing language ID systems cover a fraction of the continent's languages and mix up close relatives, and the data behind them leans heavily on Bible translations. That matters because everything downstream, from translation to speech, starts with knowing what language you are looking at.
AFROSCOPE covers 640 languages across 8 families and 7 scripts. We deduplicated and rebalanced the data so that Bible text makes up about half of it rather than three quarters or more, and we checked for leakage: zero overlap with all six evaluation benchmarks, where a widely used resource overlaps by up to 14%.
On top of the data sit a family of fine-tuned models and MIRROR, a second pass that pulls apart the languages a base model confuses.
The results: 97.9 macro-F1 on our test set against 71.9 for GlotLID-M, and the top score on 5 of 7 evaluation sets. Frontier LLMs manage 47.4 on the same task. The finding I keep coming back to is that script and close relatedness, not language family, are what predict whether one language helps another.
Come find Sang and myself at:
📍 Poster Session F, Hall H,
🕐 Today/Tuesday Oct 27, 09:00 to 10:30.
Led by Sang Yun Kwon with @arelmadany at #UBC.
This work was partially supported by CLEAR Global via the Gates Foundation. Many thanks to Christian Resch and Howard Lakougna for related discussions.
📄 Paper: lnkd.in/gW9QMZpx
💻 Code and models: lnkd.in/gW-4BPMu
#NLProc #AfricanNLP #LanguageIdentification #UBC
Today at #EMNLP2026 in Budapest, we present DRFLOW, a deep research benchmark for personalized workflow prediction.
Deep research systems are good at writing reports, but many enterprise tasks need something else: an ordered list of steps that fits one specific person. Think, for example, which forms does this employee file, in what order, and which branch of the company process applies to their case? Answering that means digging through policies, emails, and chats, ignoring distractors, and committing to actions rather than prose. This is where DRFLOW comes in.
As a benchmark, DRFLOW has 100 tasks across B2B, B2C, education, healthcare, and legal settings, with 1,246 reference steps grounded in more than 3,900 sources and seven diagnostic metrics. A mixed split adds artifacts from other questions, which makes the tasks much harder.
We also introduce DRFA, a workflow-oriented reference agent. Averaged over four LLM backbones, it beats a strong deep research baseline by about 10 F1 points and nearly doubles condition resolution, yet the best configuration (GPT-5.2) reaches only 63.9 F1, and scores drop further on the mixed split. Both agents stay highly factual. The gap is in deciding which steps apply to this user.
Come by:
📍 Poster Session F, Hall H
🕐 Today (Tuesday, Oct 27), 09:00 to 10:30
Work by our PhD student @tawkat97, with Raymond Li, Laks V.S. Lakshmanan, and Issam Laradji. A ServiceNow AI Research and UBC collaboration.
📄 Paper: lnkd.in/gyTQisBH
#NLProc #DeepResearch #LLMAgents #UBC
Today at #EMNLP2026, I will be presenting TEMPO, a benchmark for retrieval that has to reason about time.
Real questions often ask "How has X changed since 2017?" rather than "When did X happen?" Across 1,730 such questions in 13 domains, the best of 16 retrievers reaches only 32.0 NDCG@10. In RAG, answering with no retrieval at all even scored higher than with the retrievers we tested.
📍 Sunday, Oct 25, 11:00 to 12:30, Room G7-8 (Session 2)
Work led by Abdelrahman Abdallah, Mohammed Ali, and Adam Jatowt (University of Innsbruck).
📄 aclanthology.org/2026.emnlp-…
#NLProc #InformationRetrieval #RAG
UBC NLP will have a strong footprint in Budapest, with 21 papers at #EMNLP2026! 🎉
10 Main Conference, 8 Findings, 1 TACL, and 2 workshop papers (NLLP and REALM).
The work covers a lot of ground:
🌍 Culture and values: whether LLMs actually apply the cultural context they infer, South Asian meme and humor understanding, human value geometry in steering space, and value drift during post-training
🗣️ Language diversity: dialectal Arabic speech recognition (Bulbul), the linguistic landscape of Africa (AFROSCOPE), and LLM agents as computational typologists
🔎 Retrieval and RAG: temporal reasoning (TEMPO), biomedical constraints (OntologyBench), paragraph-level and long-document retrieval, and RAG on Canadian case law
🤖 Agents and reasoning: financial and numerical reasoning (MoCA-Agent), computer-use agents, deep research agents, inference as tree search, and claim verification over tabular data
📝 And more: output homogeneity in base models, discharge summarization, and large-scale authorship attribution
Huge congratulations to all the students, postdocs, and collaborators behind these papers, and to my colleagues across UBC.
If you're in Budapest, come find us at the talks and posters. And stay for ArabicNLP 2026, co-located with EMNLP!
Tomorrow, July 7, we will present our large-scale Alexandria project at ACL 2026 in San Diego!
🧬Alexandria: A Multi-Domain Dialectal Arabic Machine Translation Dataset for Culturally Inclusive and Linguistically Diverse LLMs
aclanthology.org/2026.acl-lo…
📍 Poster Session D
🕚 11:00 AM
📌 Panel C-57
🏛️ Coronado Ballroom, Level 4
📅 July 7, 2026
Alexandria is a large-scale, community-driven dataset for dialectal Arabic machine translation, designed to support culturally inclusive and linguistically diverse AI. It covers multiple Arab countries, domains, local varieties, and speaker–addressee configurations, helping move Arabic NLP beyond coarse dialect labels and toward more grounded evaluation of real language variation.
Looking forward to discussing dialectal Arabic MT, culturally grounded evaluation, and how we can build more efficient and diverse AI.
Abdellah EL Mekki, Ph.D., Samar M. Magdy, Houdaifa Atou, Ruwa AbuHweidi, Baraah Qawasmeh, OMER NACAR, Thikra Al-hibiri, Razan Saadie, Hamzah Alsayadi, Nadia Ghezaiel Hammouda, Alshima Mohammed Alkhazimi, Aya Hamod, Al-Yas Yaqoob Al-Ghafri, Wesam Ali El-Sayed, Asila Ismail al Sharji, Mohamad Ballout, Anas Belfathi, Karim Ghaddar, Serry Sibaee, Alaa Aoun, Aeej Mohammed Aseri, Lina Abureesh, Ahlam Bashiti, Majdal Yousef, Abdulaziz Hafiz, Yehdih Mohamed, Emira Hamedtou, Brakehe Emehah, Rahaf Alhamouri, Youssef Nafea, Aya El Aatar, Walid Al-Dhabyani, Emhemed S. Hamed, Sara Shatnawi, Fakhraddin Alwajih, Khalid Elkhidir, Ashwag Alasmari, Abdurrahman Gerrio, Omar Said Alshahri, AbdelRahim Elmadany, Ismail Berrada, Amir Azad Adil Al-kathiri, Fadi Zaraket, Mustafa Jarrar, Yahya Mohamed Elhadj, Hassan Alhuzali, and Muhammad Abdul-Mageed
#ACL2026 #NLProc #ArabicNLP #MachineTranslation #LLMs #MultilingualNLP #ResponsibleAI #EfficientAI
🤖 Made with AI
Excited to share our presentation today, July 6, at ACL 2026 in San Diego!
We will be presenting live at Panel C-98 in the Coronado Ballroom, Level 4.
LQM: Linguistically Motivated Multidimensional Quality Metrics for Machine Translation
Samar M. Magdy, Fakhraddin Alwajih, Abdellah El Mekki, Wesam El Sayed, Muhammad Abdul-Mageed
aclanthology.org/2026.findin…
📷 Poster Session D
📷 9:00 AM
Panel C-98, Coronado Ballroom in Level 4
📷 July 6, 2026
Looking forward to discussing linguistically grounded evaluation for machine translation and how we can move toward more efficient and diverse AI.
#ACL2026 #NLProc #MachineTranslation #ArabicNLP #Evaluation #ResponsibleAI #EfficientAI
Excited to share our July 6 presentation at ACL 2026 in San Diego!
We will be presenting:
LQM: Linguistically Motivated Multidimensional Quality Metrics for Machine Translation
Samar M. Magdy, Fakhraddin Alwajih, Abdellah El Mekki, Wesam El Sayed, Muhammad Abdul-Mageed
aclanthology.org/2026.findin…
📍 Poster Session D
🕚 11:00 AM
📅 July 6, 2026
Looking forward to discussing linguistically grounded evaluation for machine translation and how we can move toward more efficient and diverse AI.
#ACL2026 #NLProc #MachineTranslation #ArabicNLP #Evaluation #ResponsibleAI #EfficientAI
🤖 Made with AI
Join us tomorrow, July 5 for the following presentations at ACL 2026:
LQM: Linguistically Motivated Multidimensional Quality Metrics for Machine Translation
aclanthology.org/2026.findin…
Distilling the Essence: Efficient Reasoning Distillation via Sequence Truncation
aclanthology.org/2026.findin…
Zero-Shot Context-Aware ASR for Diverse Arabic Varieties
aclanthology.org/2026.findin…
See you tomorrow at ACL 2026 in San Diego!
🤖 Made with AI
🌍 We're heading to San Diego to present five papers at ACL 2026.
This year’s work from our group and collaborators spans dialectal Arabic machine translation, speech technology, machine translation evaluation, context-aware ASR, and efficient reasoning distillation. I am especially grateful to the many students, collaborators, and community contributors whose work made these papers possible.
1. Alexandria: A Multi-Domain Dialectal Arabic Machine Translation Dataset for Culturally Inclusive and Linguistically Diverse LLMs
Authors: Abdellah EL Mekki (Abdellah EL Mekki, Ph.D.), Samar M. Magdy (Samar M. Magdy), Houdaifa Atou (Houdaifa Atou), Ruwa AbuHweidi (Ruwa AbuHweidi), Baraah Qawasmeh, OMER NACAR, Thikra Al-hibiri, Razan Saadie, Hamzah Alsayadi, Dr. Eng. Nadia GHEZAIEL HAMMOUDA, Alshima Mohammed Alkhazimi, Aya Hamod, Al-Yas Yaqoob Al-Ghafri, Wesam Ali El-Sayed, Asila Ismail al Sharji, Mohamad Ballout, Anas Belfathi, Karim Ghaddar, Serry Sibaee, Alaa Aoun, Aeej Mohammed Aseri, Lina Abureesh, Ahlam Bashiti, Majdal Yousef, Abdulaziz Hafiz, Yehdih Mohamed, Emira Hamedtou, Brakehe Emehah, Rahaf Alhamouri, Youssef Nafea, Aya El Aatar, Walid Al-DHabyani, Emhemed Hamed, Sara Shatnawi, Fakhraddin Alwajih, Khalid Elkhidir, Ashwag Alasmari, Abdurrahman Gerrio, Omar Said Alshahri, AbdelRahim Elmadany, Ismail Berrada, Amir Azad Adli Al-kathiri, Fadi Zaraket, Mustafa Jarrar, Yahya Mohamed Elhadj, Hassan Alhuzali, Muhammad Abdul-Mageed
URL: lnkd.in/gfhiKBjR
2. LQM: Linguistically Motivated Multidimensional Quality Metrics for Machine Translation
Authors: Samar M. Magdy, Fakhraddin Alwajih, Abdellah El Mekki, Wesam El Sayed, Muhammad Abdul-Mageed
URL: lnkd.in/g6rz7gui
3. Zero-Shot Context-Aware ASR for Diverse Arabic Varieties
Authors: Bashar Talafha, Amin Abu Alhassan, Muhammad Abdul-Mageed
URL: lnkd.in/gKGyGQgn
4. Arab Voices: Mapping Standard and Dialectal Arabic Speech Technology
Authors: Peter Sullivan, AbdelRahim Elmadany, Alcides Alcoba Inciarte, Muhammad Abdul-Mageed
URL: lnkd.in/gMzYhT8x
5. Distilling the Essence: Efficient Reasoning Distillation via Sequence Truncation
Authors: Wei-Rui Chen, Vignesh Kothapalli, Ata Fatahibaarzi, Hejian Sang, Shao Tang, Qingquan Song, Zhipeng Wang, PhD, Muhammad Abdul-Mageed
URL: lnkd.in/ghYp5S_u
Congratulations to all co-authors and collaborators. Looking forward to discussing these works with the community at ACL 2026!
#ACL2026 #NLProc #ArabicNLP #SpeechTechnology #MachineTranslation #LLMs #ResponsibleAI #MultilingualNLP
🙌 Machine translation is an important technology that is becoming ever more pervasive in business and daily life. It is widely known that machine translation errors can be embarrassing, if not catastrophic at times. When a model makes an error, it is not just about using a word or phrase that it should not have used. Rather, the error may reflect a broader blindness to language variation, linguistic structures, cultural norms, situational nuances, and more.
For a long time, it has been difficult to characterize machine translation errors methodically and in ways that adhere to linguistic theory. Not anymore.
Today, we introduce LQM, a linguistically motivated framework for characterizing machine translation quality. Along with the framework, we construct a bidirectional parallel corpus of machine translation errors, covering seven major Arabic dialects and labeled according to our taxonomy. The corpus is derived from conversational, culturally rich content.
We use the corpus to evaluate the machine translation ability of six different LLMs against expert span-level human annotations, producing 6,113 labeled error spans across 3,495 unique erroneous sentences, along with severity-weighted quality scores.
Our accompanying paper will appear in ACL 2026 Findings.
Read the paper: “LQM: Linguistically Motivated Multidimensional Quality Metrics for Machine Translation”
Link: aclanthology.org/2026.findin…
Work by Samar M. Magdy, Fakhraddin AlWajih, Abdellah EL Mekki, and Wesam Ali El-Sayed.
#MachineTranslation #MachineTranslationErrors #MTEvaluation #ArabicNLP #ArabicDialects #LLMs #MultilingualNLP #ComputationalLinguistics #NLProc
