Mo El-Haj

VinUniversity · Vietnam

Unclaimed · AI-structured profileOpenAlex

VinUniversity NLP researcher advancing Arabic and Vietnamese language technology via shared tasks and cross-lingual QA

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-13) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.

Research profile

Output—

Not scored: the underlying OpenAlex record lacks reliable per-work volume data for this identity, so output cannot be assessed. This axis would measure total publications, first-author share, and yearly productivity.

Impact—

Not scored: citation counts and field-normalized influence are unavailable for this identity, so impact cannot be estimated. This axis would measure citations, uptake of shared-task resources, and downstream reuse of the benchmarks reviewed here.

Focus0/100

Scored 0 because the field distribution is perfectly flat: Arabic NLP, legal NLP with cross-lingual retrieval, and machine translation each hold exactly one third (0.333) of the record, while five topic clusters taper from 15 to 2. This axis measures concentration; the flat spread reads as deliberate multi-directionality — typical of a shared-task organizer — rather than drift.

Collaboration58/100

Scoring 58, this axis reflects a moderately sized, dispersed network: 17 distinct coauthors appear in the raw data, with recurring partners including Saad Ezzini, Mustafa Jarrar, Abdellah El Mekki, Paul Rayson, and Dawn Knight. The pattern suggests project-based team assembly spanning Arabic NLP and Vietnam-based legal NLP work rather than one deep bilateral partnership.

Activity93/100

The strongest axis at 93. countsByYear shows 3 works in 2024, 2 in 2025, and 18 in 2026, totalling 23 works over the last three years; all three Evolve-reviewed papers are 2026 outputs. Topic counts — Topic Modeling (15) and Natural Language Processing Techniques (14) — confirm sustained parallel workstreams, though an h-index of 1 shows volume currently outpaces accumulated impact.

Reproducibility55/100

Scoring 55, the mean over three Evolve-reviewed papers, each rated medium readiness. The verdicts consistently note sound design with honest reporting: definition-guided zero-shot classification with thin gold-label validation, honest null results on a tiny non-expert evaluation pool, and credible shared-task findings weakened by a transductive winner and single-reference spBLEU. Replicable task design, under-validated evaluation.

Research summary

Mo El-Haj is a natural language processing researcher at VinUniversity (Vietnam) whose published work spans three equally weighted strands: Arabic NLP, legal NLP with cross-lingual information retrieval, and machine translation — the bibliometric field distribution assigns exactly one third (0.333) to each. The Evolve-reviewed catalog centers on 2026: as first author, El-Haj leads AraGenre 2026, a hierarchical definition-guided shared task on Arabic genre classification, and co-authors the AlexandriaX 2026 overview, the first shared task on dialectal Arabic machine translation, plus a study of cross-lingual legal question answering for Vietnamese labour law combining retrieval, translation, and verifier-guided correction. Topic data reinforces this breadth: Topic Modeling (15) and Natural Language Processing Techniques (14) lead, with smaller clusters in sentiment analysis (4) and hate speech detection (3). Activity is the standout axis at 93 — 23 works in the last three years, jumping from 3 (2024) and 2 (2025) to 18 (2026). Collaboration scores 58 across 17 distinct coauthors, including Saad Ezzini, Mustafa Jarrar, Abdellah El Mekki, Paul Rayson, and Dawn Knight. Output and impact are unscored for lack of citation data; the recorded h-index of 1 and i10-index of 0, together with a reproducibility mean of 55 across three medium-readiness reviews, mark an early-stage profile expanding very quickly.

Collaboration pattern

The raw record lists 17 distinct coauthors, and the recurring-partner list appears to span three circles: an Arabic NLP cluster (Mustafa Jarrar, Abdellah El Mekki, AbdelRahim A. Elmadany, Samar M. Magdy, Shadi Abudalfa), Vietnam-based colleagues (Nguyen Minh Chi, Nguyen Minh Quan, Nguyen Ha Thanh, Saad Ezzini), and longer-standing corpus-linguistics links (Dawn Knight, Paul Rayson). Collaboration is broad and project-based, consistent with organizing shared tasks across communities rather than concentrating on one bilateral team.

Research keywords

Arabic NLPgenre classificationshared tasksdialectal Arabic machine translationzero-shot generalisationhierarchical classificationcross-lingual question answeringretrieval-augmented generationlegal NLPhallucination verificationtopic modelingquality estimation

Highlights

First-authoring AraGenre 2026, a definition-guided Arabic genre classification shared task
As first author, El-Haj leads a hierarchical shared task that tests zero-shot label generalisation in Arabic genre classification through definition-guided learning. Evolve judges the design sound and the results honestly reported, while flagging that gold-label validation remains thin; reproducibility readiness is rated medium.
Co-authoring AlexandriaX 2026, the first shared task on dialectal Arabic machine translation
The overview paper defines a broad three-subtask benchmark for dialectal Arabic machine translation, including MT error detection and quality estimation. Evolve finds the findings credible but cautions that a transductive S2 winner and single-reference spBLEU ranking weaken the headline claims; readiness is medium.
Bringing retrieval and verifier-guided correction to Vietnamese legal QA
This study reports that dense retrieval beats sparse retrieval for Vietnamese labour law questions and that a verifier corrects only citation errors, evaluated with faithfulness diagnostics. Evolve credits the honest null results but notes the human evaluation pool is tiny and non-expert; readiness is medium.

Timeline

2026Publishes AraGenre 2026 as first author — a hierarchical definition-guided Arabic genre classification shared task.
2026Co-authors the AlexandriaX 2026 overview, the first shared task on dialectal Arabic machine translation.
2026Contributes to cross-lingual legal question answering for Vietnamese labour law, combining retrieval, translation, and verifier-guided correction.

Papers in the Evolve catalog

Frequent co-authors

Saad Ezzini×2 Mustafa Jarrar×2 Nguyen Minh Chi×2 Shadi Abudalfa×1 Nguyen Minh Quan×1 Nguyen Ha Thanh×1 Dawn Knight×1 Paul Rayson×1 Abdellah El Mekki×1 AbdelRahim A. Elmadany×1 Samar M. Magdy×1 Saad Ezzini×1

About this profile

This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.

Bibliometric indicators (h-index 1, i10-index 0, countsByYear, topics) come from an OpenAlex identity snapshot; no ORCID is linked, so this profile is name-resolved from public bibliographic data and namesake merging or splitting is possible.
Output and impact dimensions are unscored because reliable citation and volume data are missing for this identity.
Only three papers — all 2026 arXiv preprints — have been reviewed by Evolve; venue-reviewed publications are not represented in the catalog.
The reproducibility score (mean 55) averages three medium-readiness reviews and is not a validated replication record.