Jisoo Kim (Sungkyunkwan University)
Sungkyunkwan University · South Korea
Neuro-symbolic compliance: turning safety rules and regulations into executable AI checks (Sungkyunkwan University)
2
papers here
2
reviewed by Evolve
2
first-author
0
corresponding
—
works · OpenAlex
0
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 scored: OpenAlex indexes only two works, both appearing in 2026, with no h-index or i10-index signal yet. This axis would measure total publication volume, venue spread, and sustained productivity once a longer publication record accumulates.
Impact—
Not yet scored: with h-index 0 and i10-index 0 across just two 2026 preprints, citation impact cannot yet be measured meaningfully. This axis would track citations, citation velocity, and uptake of these two compliance papers by follow-up work.
Focus0/100
Scores 0/100: the OpenAlex field distribution splits exactly 50/50 between AI safety and multimodal machine learning and AI for legal compliance (NLP / neuro-symbolic reasoning), and topics scatter across five clusters — explainable AI (×2), topic modeling, advanced neural network applications, safety systems engineering in autonomy, and AI in law. At this two-paper stage the record reads as two parallel compliance threads rather than one concentrated line.
Collaboration39/100
Scores 39/100, computed from 6 distinct coauthors. Both cataloged papers carry first-author (non-corresponding) positions, and the recurring partners — TaeYoon Kwack, Jinwoo Jang, Honguk Woo, and Woo Kyung Kim — are all listed with Sungkyunkwan University, indicating a compact, single-institution team rather than an external or international collaboration network so far.
Activity32/100
Scores 32/100. The counts-by-year record shows 2 works, both in 2026, and recent3yWorks is likewise 2 — an active debut year but a very short window. Both outputs are arXiv preprints with Kim as first author, so current activity reflects a concentrated 2026 burst rather than sustained multi-year publishing.
Reproducibility38/100
Scores 38/100, the mean repro-readiness across 2 reviewed papers. GUARDEN (Visual Compliance via Executable Safety Rule Entailment) is rated medium, while Code-as-Auditor is rated low, with Evolve citing LLM-judge circularity and GPT-5-derived data as the limits on independent replication of the reported +12.3 PIPA gain.
Research summary
Jisoo Kim is an early-stage researcher at Sungkyunkwan University, South Korea, whose work makes compliance machine-checkable — turning safety rules and legal regulations into executable logic that AI systems can be audited against. The Evolve catalog lists two 2026 arXiv preprints, both first-authored by Kim. The first, GUARDEN, compiles safety rules into executable, scene-graph-grounded logic for visual compliance and vision-language model safeguarding; Evolve judged its benchmark gains impressive but found external validity and statistical rigor unproven. The second, Code-as-Auditor, converts regulations into code for automated compliance checking with self-verified checklists, reporting a +12.3 F1 gain on PIPA — gains tempered by LLM-judge circularity and GPT-5-generated data. The bibliometric record mirrors this dual track: OpenAlex splits the field distribution evenly (0.5/0.5) between AI safety and multimodal machine learning and AI for legal compliance, with explainable AI the most frequent topic (2 papers). Citation indicators remain at baseline (h-index 0, i10-index 0), with two works indexed in the past three years, consistent with a 2026 debut. Collaboration is compact — six distinct coauthors, largely Sungkyunkwan colleagues such as TaeYoon Kwack, Jinwoo Jang, Honguk Woo, and Woo Kyung Kim. This name-resolved profile is built from OpenAlex records and Evolve reviews; read it as an early-stage snapshot.
Collaboration pattern
Collaboration is compact and single-institution: 6 distinct coauthors are recorded, and the frequent partners — TaeYoon Kwack, Jinwoo Jang, Honguk Woo, and Woo Kyung Kim — are all listed with Sungkyunkwan University. Kim takes first-author (non-corresponding) position on both cataloged papers, consistent with a hands-on contributor inside a stable SKKU team rather than a distributed or international network (collaboration dimension: 39/100).
Research keywords
Highlights
GUARDEN: compiling safety rules into executable, scene-graph-grounded checks
Code-as-Auditor: regulation-to-code compliance checking
Timeline
- 2026Publishes the first-authored arXiv preprint 'Visual Compliance via Executable Safety Rule Entailment' (GUARDEN), entering AI safety and multimodal machine learning.
- 2026Publishes the second first-authored arXiv preprint, 'Code-as-Auditor: Executable Compliance Reasoning via Regulation-to-Code', opening the AI-for-legal-compliance thread.
- 2026OpenAlex records both works within a single year: countsByYear shows 2 publications in 2026, with h-index and i10-index at 0 — the starting point of this profile's record.
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.
- Bibliometrics are derived from OpenAlex only (identity: openalex); no ORCID is linked, and 'Jisoo Kim' is a common Korean name, so this profile is name-resolved from public bibliographic data and may merge or split identities.
- Output and impact dimensions are unscored: only two works (both 2026) are indexed, with h-index 0 and i10-index 0 — too early for meaningful productivity or citation measures.
- The qualitative assessment rests on just two Evolve-reviewed arXiv preprints from 2026; no peer-reviewed venue records are present in the provided data.
- The reproducibility score (mean 38) is averaged over only 2 papers, so the small sample limits its reliability.