Klagenfurt researcher: governed LLM decision support and efficient language models; first-author, active early-stage.
2
papers here
2
reviewed by Evolve
2
first-author
0
corresponding
—
works · OpenAlex
4
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-15) 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—
The output axis is not scored yet: this profile lacks a verified total-publication count and venue breakdown. Its only available signal is recent3yWorks = 15 over the last three years, too incomplete to aggregate into a reliable productivity score.
Impact—
Citation-impact data is insufficient for scoring: the record shows only h-index 4 and i10-index 1. This axis would measure aggregate citations, citation trajectory and field-level influence once enough citable works accumulate.
Focus0/100
Focus scores 0, reflecting deliberate breadth rather than narrow specialisation: field distribution splits evenly (0.5/0.5) between AI governance/trustworthy LLM systems and efficient language models, while OpenAlex topic counts span Advanced Neural Network Applications (7), Digital Transformation in Industry (7), Semantic Web and Ontologies (6), Handwritten Text Recognition (6) and Topic Modeling (4). The two 2026 preprints indicate the current centre of gravity is governed, auditable LLM systems.
Collaboration36/100
Collaboration scores 36, a compact network: the record lists only 5 distinct coauthors, with Vahid Tavakkoli and Kyandoghere Kyamakya recurring most often. Both Evolve-reviewed papers are first-authored (neither corresponding), consistent with a small stable working group where the researcher leads writing and experiments rather than coordinating a wide external network.
Activity81/100
Activity scores 81, the strongest axis: recent3yWorks is 15, and the counts-by-year series climbs to its maximum of 10 works in 2026, after 7 in 2020, 1 in 2021, 2 in 2022 and 5 in 2024. Both first-author arXiv preprints appeared in 2026, indicating a sustained and currently accelerating publishing cadence.
Reproducibility78/100
Reproducibility scores 78, computed as the mean readiness rating across 2 reviewed papers (reproMean 78). Policy-as-Skill is rated 'high' readiness; TinyCeNN-LM 'medium'. Evolve's verdicts on both independently stress honest reporting — the governance paper's gains are tied to a degenerate subset with synthetic, single-expert labels, and the conversion paper's claims are modestly scaled — a pattern that supports the score without overstating the evidence.
Research summary
Kabeh Mohsenzadegan is an early-stage researcher at the University of Klagenfurt (Austria) whose OpenAlex-resolved profile shows a first-author-driven output pattern. Two 2026 arXiv preprints reviewed by Evolve define the current agenda. The first, Policy-as-Skill, packages LLM governance as versioned executable skills with evidence grounding, deterministic control and audit; it earns the 'high' reproducibility-readiness rating, though the verdict notes headline gains ride a degenerate subset, a synthetic corpus and single-expert labels. The second, TinyCeNN-LM, proposes quality-gated conversion of pretrained attention into CeNN-inspired cellular-recurrent layers with accept/rollback, rated 'medium' and judged honest but small-scale, with no measured speedup. Around these papers, the bibliometric record shows an activity score of 81, with 15 works over the last three years and yearly counts peaking at 10 works in 2026 (7 in 2020, 1 in 2021, 2 in 2022, 5 in 2024). Field distribution splits evenly between AI governance/trustworthy LLM systems and efficient language models, and OpenAlex topics span neural network applications, industrial digital transformation, semantic web and ontologies, handwritten text recognition and topic modeling — a breadth reflected in a focus score of 0. Citation signals remain thin (h-index 4, i10-index 1), so the output and impact axes are not yet scored. Collaboration is compact (score 36; 5 distinct coauthors), anchored by repeat partners Vahid Tavakkoli and Kyandoghere Kyamakya.
Collaboration pattern
Collaboration is compact: only 5 distinct coauthors appear in the OpenAlex record (collaboration score 36). Vahid Tavakkoli and Kyandoghere Kyamakya recur as the primary partners, consistent with a stable, small working group. Both 2026 papers are first-authored but not corresponding, suggesting the researcher leads the writing and core experiments within a fixed small team rather than acting as an independent corresponding author or joining large consortia.
Research keywords
Highlights
Policy-as-Skill: LLM governance packaged as versioned executable skills
TinyCeNN-LM: quality-gated attention-to-recurrent conversion
A reproducibility-first, first-author track record
Both catalog-reviewed papers are first-authored (neither corresponding) and jointly yield a reproducibility score of 78 (reproMean over 2 papers). Evolve's verdicts repeatedly flag honest reporting even where evidence is thin — a consistent pattern across the governance and efficiency tracks.
Timeline
- 2020Earliest year in the OpenAlex record: 7 works indexed, the largest pre-2026 annual total in the snapshot.
- 2022Only 2 works indexed — a quiet stretch between the 2020 peak and the later ramp-up.
- 20245 works indexed as annual output begins climbing again.
- 2026Most active year on record: 10 works indexed, including both Evolve-reviewed first-author arXiv preprints (Policy-as-Skill and TinyCeNN-LM).
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.
- Identity and bibliometrics are resolved from a public OpenAlex record; publications by same-name or similar-name authors may be merged or split.
- Output and impact axes are unscored: only h-index (4), i10-index (1) and recent3yWorks (15) are available as sparse signals.
- Both reviewed papers are 2026 arXiv preprints; Evolve verdicts are independent reviews, not evidence of peer-reviewed venue acceptance.
- Counts-by-year has gaps (no entries for 2023 and 2025), so year coverage in the record may be incomplete.