TaeYoon Kwack (Sungkyunkwan University)
Sungkyunkwan University · South Korea
Early-career researcher linking executable AI safety rules and legal-compliance reasoning at Sungkyunkwan University
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
1
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-14) 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 scorable: the underlying data — cumulative publication count and venue mix beyond the four recently indexed works — is insufficient in the current record. This axis would measure total and per-year research production as the early-career publication record fills in.
Impact—
Not yet scorable: citation evidence is too thin — an h-index of 1 and an i10-index of 0 cannot support a meaningful impact estimate. This axis would measure citation uptake, influence on follow-up work, and scholarly reach.
Focus0/100
The focus score is 0/100 because output splits exactly evenly (0.5/0.5) between two distinct fields: AI safety with multimodal machine learning, and AI for legal compliance via NLP and neuro-symbolic reasoning. Topic counts reinforce the spread — topic modeling (2) and explainable AI (2) sit alongside single-count robotics learning topics. The unifying thread is executable-rule methodology, but concentration around one field has not yet formed.
Collaboration39/100
The collaboration score is 39/100, based on 6 distinct coauthors — a small, concentrated network. Frequent collaborators Jisoo Kim, Jinwoo Jang, and Honguk Woo are all at Sungkyunkwan University, with Woo Kyung Kim also recurring. On both 2026 catalog papers Kwack holds a middle-author, non-corresponding position, indicating contributor roles inside a stable single-institution team rather than an independent or cross-institution collaboration pattern so far.
Activity47/100
The activity score is 47/100. The indexed record is entirely recent: 2 works in 2025 and 2 in 2026, giving 4 works over the last three years, all released as arXiv preprints. This is a steady but early-career cadence — volume is still low and there is no longer publication history to evidence sustained output.
Reproducibility38/100
The reproducibility score is 38/100, averaged over 2 reviewed papers. "Visual Compliance via Executable Safety Rule Entailment" is rated medium readiness, with gains shown on its own benchmark while external validity and statistical rigor remain unproven. "Code-as-Auditor: Executable Compliance Reasoning via Regulation-to-Code" is rated low, reflecting LLM-judge circularity and dependence on GPT-5-generated data.
Research summary
TaeYoon Kwack is an early-career researcher at Sungkyunkwan University (South Korea) whose work centers on making AI systems verifiable against explicit human-written rules. The bibliometric record, resolved from OpenAlex, shows an even split (0.5/0.5) between two fields: AI safety and multimodal machine learning, and AI for legal compliance built on NLP and neuro-symbolic reasoning. Two 2026 arXiv preprints anchor this dual agenda. GUARDEN ("Visual Compliance via Executable Safety Rule Entailment") compiles safety rules into executable logic grounded in scene graphs, earning a medium reproducibility rating with strong gains on its own benchmark but unproven external validity. "Code-as-Auditor" converts regulations such as GDPR and PIPA into executable compliance checks with self-verified checklists, reporting a +12.3 F1 gain on PIPA, though LLM-judge circularity and GPT-5-derived data temper the claims and reproducibility is rated low. Topic signals — topic modeling (2), explainable AI (2), plus robotics learning topics — suggest breadth beyond the two catalogued papers. Output indicators remain modest and very recent: 4 works in the last three years (2 in 2025, 2 in 2026), an h-index of 1, and an i10-index of 0. Collaboration is concentrated: 6 distinct coauthors, predominantly at Sungkyunkwan University, with Kwack in middle-author, non-corresponding roles on both catalogued papers.
Collaboration pattern
Collaboration is small and single-institution: 6 distinct coauthors in total, with the recurring cluster — Jisoo Kim, Jinwoo Jang, and Honguk Woo — all at Sungkyunkwan University, plus repeat collaborator Woo Kyung Kim. Kwack holds middle-author, non-corresponding positions on both 2026 catalog papers, consistent with contributing within an established local team (collaboration score: 39/100). No cross-institution or industry partners appear in the available record.
Research keywords
Highlights
Executable safety-rule entailment for visual compliance (GUARDEN)
Regulation-to-code compliance checking (Code-as-Auditor)
A shared "rules become executable" methodology
Across both 2026 preprints and both half-weight fields, the constant is turning human-written rules — safety rules or legal regulations — into executable, checkable artifacts, flanked by topic-modeling and explainable-AI signals in the topic profile.
Timeline
- 2025First indexed works appear: 2 publications recorded in 2025 in the OpenAlex-derived counts-by-year data.
- 2026Co-authored "Visual Compliance via Executable Safety Rule Entailment" (arXiv preprint, middle-author position).
- 2026Co-authored "Code-as-Auditor: Executable Compliance Reasoning via Regulation-to-Code" (arXiv preprint, middle-author position).
- 2026Recent-3-year output reaches 4 works (2 in 2025, 2 in 2026), with an h-index of 1 and an i10-index of 0.
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 is resolved from an OpenAlex record with no linked ORCID; name-based resolution carries the usual risk of merged or split records for this name.
- Bibliometrics come from OpenAlex only; the output and impact axes are unscored because citation and venue data are too sparse (h-index 1, i10-index 0).
- Both cataloged papers are 2026 arXiv preprints with middle-author, non-corresponding positions; findings are not yet peer-review validated, and Evolve has reviewed only 2 papers for this researcher.
- Affiliation details for the frequent-coauthor list are only partially specified (Woo Kyung Kim's affiliation is not given).