Felermino D. M. A. Ali

Lúrio University · Mozambique

Unclaimed · AI-structured profileOpenAlex

NLP for Mozambican Bantu languages: machine translation evaluation and culturally grounded AI assessment

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-17) 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—

Output scoring is not yet available: beyond the 3 works recorded in the recent three-year window, the OpenAlex snapshot lacks the aggregated publication-volume data this axis needs, so production volume and its trend cannot be scored.

Impact—

Impact scoring is not yet available: the snapshot records h-index 0 and i10-index 0 with no citation distribution, so this axis, which would measure citation uptake and influence of the work, remains unscored.

Focus100/100

The focus score is 100, the strongest signal in this profile. Topic data concentrates on Topic Modeling and Natural Language Processing Techniques (two works each), with single works in Speech Recognition and Synthesis, Ethics and Social Impacts of AI, and Algorithms and Data Compression. The one field-tagged work sits in natural language processing (machine translation), and both cataloged papers address low-resource Mozambican languages and culturally grounded evaluation of AI in African contexts.

Collaboration51/100

Collaboration scores 51, a mid-range value consistent with 12 distinct coauthors in the recent window. The frequent-coauthor list splits into Mozambican colleagues (Delfina Lázaro Mateus, Manuel Valente Mangue) and a larger pan-African and international group including Millicent Ochieng, Elizabeth A. Ankrah, Aditya Vashistha, Anja Thieme and Jacki O'Neill, which matches the cross-community human-centered evaluation paper. The first-author MT evaluation work pairs with the Mozambican coauthors.

Activity40/100

Activity scores 40. The counts-by-year record contains a single year, 2026, with 3 works, and recent3yWorks is likewise 3; no earlier years appear in the snapshot. This suggests a newly visible or early-stage publishing record, where all cataloged output — both papers dated 2026 — falls within one annual window, so longer-run activity rhythms cannot yet be established.

Reproducibility55/100

Reproducibility scores 55, computed as reproMean from a single assessed paper (reproPapers 1). That paper — the first-author FLORES+ extension — carries a medium reproducibility readiness rating, consistent with benchmark construction and controlled MT evaluation experiments whose artifacts are often shareable, while Evolve has not yet assessed the second cataloged paper, so this score rests on a one-paper evidence base.

Research summary

Felermino D. M. A. Ali is a researcher at Lúrio University in Mozambique working on natural language processing for low-resource Bantu languages; this profile is name-resolved from public OpenAlex bibliographic data. The cataloged work concentrates on machine translation (MT) evaluation: a 2026 first-author arXiv study extends FLORES+ with Mozambican Xichangana, Nyanja and Sena and measures how conflating closely related language varieties inflates MT scores, finding that sibling references inflate evaluation results while controlled evidence remains convincing and fine-tuning recovers most of the loss. A second 2026 collaboration, 'Toward Cultural Alignment', applies human-centered evaluation to multimodal AI stories across five African communities, extending the agenda toward culturally grounded AI assessment. Topic data reflects this concentration: Topic Modeling and Natural Language Processing Techniques lead with two works each, alongside single works in Speech Recognition and Synthesis, Ethics and Social Impacts of AI, and Algorithms and Data Compression. The snapshot records 3 works in the recent three-year window and 12 distinct coauthors spanning Mozambican colleagues and international partners, with a collaboration score of 51 and a focus score of 100. Citation-based impact metrics are not yet meaningful — h-index and i10-index are recorded as 0 and the output and impact radar axes remain unscored. Reproducibility is scored 55 from one assessed paper with medium reproducibility readiness.

Collaboration pattern

Collaboration scores 51 with 12 distinct coauthors in the recent window. The pattern is two-tier: Mozambican colleagues Delfina Lázaro Mateus and Manuel Valente Mangue anchor the first-author MT evaluation work, while a pan-African and international group — Millicent Ochieng, Elizabeth A. Ankrah, Samuel Chege Maina, Aditya Vashistha, Anja Thieme, Jacki O'Neill — corresponds to the human-centered cultural-alignment paper. No solo-authored work appears in the cataloged record.

Research keywords

machine translation evaluationFLORES+ benchmarklow-resource languagesBantu languagesMozambican languageslanguage varietiestopic modelingspeech recognition and synthesisethics and social impacts of AIhuman-centered evaluationmultimodal AI

Highlights

Extending FLORES+ with Xichangana, Nyanja and Sena
First-author 2026 arXiv study adds three Mozambican Bantu varieties to the FLORES+ benchmark, creating controlled evaluation sets for multilingual MT in previously underrepresented languages; Evolve rates its reproducibility readiness as medium.
Quantifying the cost of variety conflation in MT evaluation
The same study shows sibling references inflate multilingual MT scores; the controlled evidence is judged convincing, and fine-tuning recovers most of the gap — a direct caution for low-resource MT benchmarking practice.
Human-centered evaluation of multimodal AI stories across five African communities
Co-authored 2026 arXiv paper on cultural alignment of multimodal AI stories with five African communities; this record has no verdict, keywords or reproducibility rating yet, and the author position is middle.

Timeline

20263 works recorded in the OpenAlex counts-by-year data, the only year present in this snapshot
2026First-author arXiv paper adds Mozambican Xichangana, Nyanja and Sena to FLORES+ and quantifies variety-conflation effects in MT evaluation
2026Co-authors the human-centered evaluation of multimodal AI stories across five African communities (middle author position)
2026Topic snapshot records Topic Modeling and Natural Language Processing Techniques as leading topics, with 12 distinct coauthors in the recent window

Papers in the Evolve catalog

2026-09-12first author
Measuring the Cost of Variety Conflation in Multilingual MT Evaluation: Adding Mozambican Xichangana, Nyanja and Sena to FLORES+

Frequent co-authors

Delfina Lázaro Mateus×1 Manuel Valente Mangue×1 Millicent Ochieng×1 Elizabeth A. Ankrah×1 Najeeb Gambo Abdulhamid×1 Migisha Boyd×1 Stephanie Nyairo×1 Mercy Muchai×1 Samuel Chege Maina×1 Aditya Vashistha×1 Anja Thieme×1 Jacki O'Neill×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.

Bibliometrics are computed from OpenAlex only; no ORCID is linked, so identity is name-resolved from public bibliographic data.
Citation-based output and impact radar axes are unscored; the snapshot records h-index and i10-index as 0.
Only two papers have been reviewed by Evolve; one lacks a verdict, keywords and a reproducibility rating.
Counts by year contain only 2026 (3 works), so longer-term publication trends cannot be assessed.