Jan Scholtes

—

Unclaimed · AI-structured profileOpenAlex

Speech and NLP researcher focused on attention analysis for machine translation and efficient zero-shot SpeechLLMs

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

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—

Not yet scored: the OpenAlex-linked snapshot lists only one 2026 work and no aggregate output volume, so this axis cannot be computed. It would measure total publication count and productivity trajectory across venues over time.

Impact—

Not yet scored: citation indicators are effectively empty (h-index 0, i10-index 0) with only one 2026 work, too little signal to grade influence. This axis would measure citation uptake, h-index trajectory and scholarly visibility.

Focus100/100

Score 100 reflects a perfectly concentrated field profile: fieldDist lists exactly one field, speech and natural language processing, with one paper. Topic assignments — Natural Language Processing Techniques, Neurobiology of Language and Bilingualism, and Multimodal Machine Learning Applications — all orbit the language-and-speech space, and both catalog papers address language technology (machine translation attribution; SpeechLLM alignment), so topical focus is genuinely narrow.

Collaboration39/100

Score 39 reflects a small but coherent network: 6 distinct co-authors in the OpenAlex-linked snapshot. The frequent-coauthor list is dominated by Gerasimos Spanakis (listed repeatedly), alongside Yusuf Can Semerci, Abderrahmane Issam and Paweł Mąka — a compact cluster consistent with shared speech/NLP lab work. Both catalog papers sit in middle author positions without corresponding-author roles, suggesting team-based contributions rather than an independent group so far.

Activity20/100

Score 20 signals low measured activity: the snapshot's countsByYear shows one publication in 2026 and recent3yWorks is 1, while the Evolve catalog holds two 2026 arXiv preprints — a gap suggesting incomplete indexing. On current bibliometric data alone, recent output is minimal and concentrated in a single year.

Reproducibility100/100

Score 100 is based on a single assessed paper: reproMean is 100 across reproPapers = 1. The reviewed SpeechLLM alignment paper carries a high reproducibility readiness rating, with DTW-aligned CIF and single-layer knowledge distillation described as cheap, implementable components; the Evolve verdict nonetheless flags that headline claims rest on single-seed runs, so this score reflects depth of method detail rather than multi-run validation.

Research summary

Jan Scholtes is a researcher in speech and natural language processing whose record is name-resolved from public OpenAlex bibliographic data. The Evolve catalog holds two 2026 arXiv papers with this researcher in a middle author position: “Scaling Attention Head Analysis via Gradient-Based Attribution in Context-Aware Machine Translation” and “Align, Integrate, and Fire: Efficient Token-Level Alignment for Zero-Shot SpeechLLMs”. The latter is the only paper with a completed Evolve review; the verdict notes that DTW-aligned CIF plus single-layer knowledge distillation enable cheap zero-shot SpeechLLMs, that gains over SpeechMapper look real, but that headline claims rest on single-seed runs. Reproducibility readiness for that paper is rated high, giving a reproducibility mean of 100 across the one assessed paper. The bibliometric snapshot shows a perfect focus score (a single field, speech and natural language processing), six distinct co-authors with Gerasimos Spanakis recurring most often alongside Yusuf Can Semerci, Abderrahmane Issam and Paweł Mąka, one work in the last three years (2026), and h-index and i10-index both at zero — the shape of an early-stage, sparse record rather than an established citation footprint. Output and impact axes are therefore not scored. Assigned topics span Natural Language Processing Techniques, Neurobiology of Language and Bilingualism, and Multimodal Machine Learning Applications.

Collaboration pattern

Co-authorship is concentrated in one small cluster: 6 distinct co-authors, with Gerasimos Spanakis recurring most often in the frequent-coauthor list, alongside Yusuf Can Semerci, Abderrahmane Issam and Paweł Mąka — a pattern consistent with a single speech/NLP research group. Both catalog papers are middle-position, non-corresponding contributions, indicating collaborative team work rather than solo or leading-author output so far.

Research keywords

SpeechLLMzero-shot speech recognitionspeech translationContinuous Integrate-and-Fire (CIF)Dynamic Time Warping alignmentknowledge distillationattention head analysisgradient-based attributioncontext-aware machine translationmultimodal machine learningnatural language processing

Highlights

Cheap zero-shot SpeechLLMs via DTW-aligned CIF and single-layer distillation
Evolve's review of “Align, Integrate, and Fire: Efficient Token-Level Alignment for Zero-Shot SpeechLLMs” (2026, arXiv) concludes that DTW-aligned Continuous Integrate-and-Fire plus single-layer knowledge distillation enable cheap zero-shot SpeechLLMs, and that gains over SpeechMapper look real — though headline claims rest on single-seed runs. Reproducibility readiness: high.
Gradient-based attribution scales attention-head analysis in context-aware MT
The 2026 arXiv preprint “Scaling Attention Head Analysis via Gradient-Based Attribution in Context-Aware Machine Translation” extends attention-head analysis to context-aware machine translation using gradient-based attribution. It is cataloged in a middle author position and has not yet been reviewed by Evolve, so no verdict or reproducibility rating exists.

Timeline

2026arXiv preprint: “Scaling Attention Head Analysis via Gradient-Based Attribution in Context-Aware Machine Translation” (middle author position, no Evolve review yet).
2026arXiv preprint: “Align, Integrate, and Fire: Efficient Token-Level Alignment for Zero-Shot SpeechLLMs” (middle author position; assigned to speech and natural language processing).
2026Evolve review completed for the SpeechLLM alignment paper with high reproducibility readiness; the OpenAlex snapshot records one indexed 2026 work, with h-index and i10-index both at 0.

Papers in the Evolve catalog

Frequent co-authors

Yusuf Can Semerci×2 Abderrahmane Issam×1 Gerasimos Spanakis×1 Gerasimos Spanakis×1 Paweł Mąka×1 Gerasimos Spanakis×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 come from OpenAlex only; the record is name-resolved from public bibliographic data, and “Jan Scholtes” is a relatively common name, so identity should be verified before relying on this profile.
Output and impact dimensions are unscored: citation coverage is minimal (h-index 0, i10-index 0 in the snapshot) and no aggregate output volume is available.
Only one of the two catalog papers has a completed Evolve review; the machine translation attribution paper has no verdict, keywords or reproducibility rating yet.
Institution, country and ORCID are absent from the input data, so affiliation context cannot be stated.