FPT University researcher in educational AI: knowledge-tracing fairness audits and sub-3B LLM essay-scoring benchmarks
2
papers here
0
reviewed by Evolve
1
first-author
0
corresponding
—
works · OpenAlex
0
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-09-28) 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—
This axis is not yet scored: the underlying bibliographic record is too thin to compute a reliable output measure. It would capture cumulative publication volume, venue quality, and year-over-year output once OpenAlex-linked works accumulate beyond the current snapshot.
Impact—
Impact is unscored because citation-based evidence is not yet available in this snapshot. The axis would measure citations, h-index trajectory, and uptake of the two 2026 preprints once the OpenAlex record accrues citing works.
Focus—
Focus is not scored here: although OpenAlex lists five topics each with a single count, the profile lacks enough works to quantify topical concentration. This axis would measure how tightly research clusters around core areas versus spreading across them.
Collaboration32/100
OpenAlex assigns a collaboration score of 32/100, built on four distinct coauthors — Nguyen Dung Son, Nguyen Huu Loi, Truong Viet Vu, and Nguyen Thai Anh — who recur across this profile. The pattern suggests a small, stable working group rather than a wide international network: both 2026 preprints sit within this compact circle. As the output record grows, the coauthor graph will show whether these ties extend outward.
Activity40/100
The activity score of 40/100 reflects a short but current publication trail: OpenAlex logs three works within the recent three-year window, with the by-year counts placing all three in 2026. The two Evolve-reviewed arXiv preprints are both 2026 releases, so visible output is recent and concentrated rather than sustained over many years.
Reproducibility—
Reproducibility has no radar score yet. The axis would assess released code, data, and setup detail across papers; Evolve's per-paper review rates both 2026 preprints as medium repro-readiness, but an aggregate axis needs more evidence than the current record provides.
Research summary
Dang Quang Minh is an early-stage researcher at FPT University (Vietnam) working at the intersection of educational artificial intelligence and responsible machine learning. Evolve's catalog includes two of his 2026 arXiv preprints. As first author, he led "Accuracy Is Not Enough: A Cross-Architecture Audit of Demographic Bias in Deep Knowledge Tracing", which Evolve characterizes as the first cross-architecture audit of demographic bias in deep knowledge tracing, applying ABROCA with careful statistics, while noting that headline p-values sit at the permutation floor. He also contributed as a middle author to a benchmark asking how far sub-3B open language models go in zero-shot essay scoring on an 8 GB consumer GPU; its candid conclusion is that the best quadratic weighted kappa (0.388), achieved with rubric-decomposed prompting, still loses to a length-only baseline. A consistent methodological signature runs through both papers: statistically careful, honest reporting of negative or sobering results, with medium repro-readiness ratings from Evolve. OpenAlex currently records an h-index of 0, three works in the recent three-year window, four distinct coauthors, and a broad five-topic profile spanning topic modeling, NLP techniques, graph algorithms, AI ethics, and stochastic gradient optimization — a picture of an emerging career just taking shape.
Collaboration pattern
OpenAlex records four distinct coauthors, and the frequent-coauthor list — Nguyen Dung Son, Nguyen Huu Loi, Truong Viet Vu, and Nguyen Thai Anh — points to a compact, recurring team behind both 2026 preprints. The collaboration score of 32/100 reflects a small network so far: the work appears team-based rather than solitary, but outreach beyond this circle is not yet visible in the data.
Research keywords
Highlights
First cross-architecture audit of demographic bias in deep knowledge tracing
Sub-3B open LLMs pushed to the limit on zero-shot essay scoring with an 8 GB GPU
A candid, statistically careful benchmarking style
Both Evolve-reviewed papers are described as statistically careful and honest — including when results disappoint — and both carry medium repro-readiness ratings, suggesting a deliberate emphasis on measurement rigor in educational AI evaluation.
Timeline
- 2026First-authored "Accuracy Is Not Enough: A Cross-Architecture Audit of Demographic Bias in Deep Knowledge Tracing" released on arXiv.
- 2026Co-authored benchmark of sub-3B open language models for zero-shot essay scoring on an 8 GB consumer GPU released on arXiv.
- 2026OpenAlex by-year counts record three works in 2026 — the entire visible recent output lands in this single year.
Papers in the Evolve catalog
Frequent co-authors
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 0, i10-index 0, work counts, topics) come from an OpenAlex identity with no ORCID link; name disambiguation may be imperfect for a common Vietnamese name.
- Coverage is limited to the two 2026 arXiv preprints reviewed by Evolve; journal publications, datasets, software, awards, and funding are outside this snapshot.
- Output, impact, focus, and reproducibility radar axes are unscored because the underlying bibliometric data is insufficient; no values have been estimated.
- Repro-readiness ratings (medium for both papers) reflect Evolve's editorial review, not an independent verification of released code or data.