Junyoung Park (Seoul National University): Korean visual speech recognition and LLM-agent social simulation
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-16) 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—
Insufficient data: no output score can be computed yet. This axis would measure cumulative publication volume and per-year productivity across indexed venues, which the current snapshot — limited to two catalogued 2026 papers — cannot support.
Impact—
Insufficient data: citation counts, citation velocity, and influence relative to field norms are absent from the snapshot. This axis would measure how widely the catalogued work is cited and built upon by other researchers.
Focus0/100
The focus score is 0, driven by an exact 50/50 split in the raw field distribution between speech processing (audio-visual speech recognition / machine learning) and computational social science (LLM agent simulation). Each catalogued 2026 paper sits in a different line with little visible methodological bridge, so activity reads as dual-track rather than concentrated in one field.
Collaboration58/100
Collaboration scores 58, computed from 17 distinct coauthors in the snapshot. Twelve names recur across the catalogued work — among them Chaemin Jang, Junsik Min, Jaewoo Choi, Donggyu Lee, Hyunwoo J. Kim, Jungwon Kim, and Juho Kim — indicating stable team settings. Park is last author on the speech paper and middle author on the simulation paper, and is not listed as corresponding author on either.
Activity—
Insufficient data: no activity score is available. This axis would measure recent publication cadence — papers per year over the last few years — but the snapshot contains only two catalogued 2026 papers, too few to establish a trend.
Reproducibility78/100
Reproducibility is the strongest measured axis at 78 (reproMean across 2 assessed papers). The LLM deliberation paper rates high readiness — strong controls and descriptive statistics support independent replication — while the Korean lipreading paper is medium: its paired evaluation is rigorous, but missing prior-method baselines make the claimed 9.95–12.19% vs 26.64 CER gap harder to verify externally.
Research summary
Junyoung Park is a researcher at Seoul National University whose catalogued work spans two distinct lines: speech processing, centered on audio-visual speech recognition and lipreading, and computational social science built on LLM agent simulation. The field distribution splits evenly at 0.5 per line, so the focus score reads 0: effort is divided rather than concentrated. On the speech side, Park is last author of a personalized Korean lipreading study on the OLKAVS dataset, framed as transfer, census and adaptation: it reports character error rates of 9.95–12.19% against a 26.64 comparison, supported by a speaker census and LoRA-based personalization. Evolve's review credits the rigorous paired evaluation but flags missing prior-method baselines. On the simulation side, Park is a middle author of a study of LLM persona deliberation, which finds a 29-percentage-point gap between simulated and surveyed opinion representation, and stance shifts occurring without genuine peer exchange; controls are strong, though the analysis stays descriptive. Reproducibility is the strongest measured dimension at 78, computed from the two assessed papers (readiness high and medium), followed by collaboration at 58 across 17 distinct coauthors. Output, impact, and activity metrics are not yet computable and are reported as missing; both catalogued papers are 2026 arXiv preprints.
Collaboration pattern
Collaboration scores 58, built on 17 distinct coauthors, 12 of whom recur across the catalogued work — including Chaemin Jang, Junsik Min, Jaewoo Choi, Donggyu Lee, Haiin Lee, Namhee Kim, Hyunwoo J. Kim, Jungwon Kim, Juho Kim, Nuri Kim, Jihee Kim, and Se Un Park. On the two papers Park is last author (speech) and middle author (simulation), never corresponding author — a contributor profile inside larger teams rather than a solo or lead-PI pattern.
Research keywords
Highlights
Claimed Korean VSR state of the art on OLKAVS
Personalization via speaker census and LoRA adaptation
LLM persona deliberation misrepresents the population
Timeline
- 2026Enters the catalog in speech processing: last-authored arXiv preprint on personalized Korean lipreading on OLKAVS, reporting 9.95–12.19% CER against a 26.64 comparison.
- 2026Publishes (middle author) an arXiv preprint in computational social science diagnosing representation and interaction failures in LLM persona deliberation, including a 29-percentage-point representation gap.
- 2026Both strands land in the same year, producing the 50/50 field split behind the focus score of 0 and forming the 2-paper reproducibility sample (mean 78).
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.
- Radar scores are computed from public bibliometric data; venue and citation coverage may be incomplete.
- No ORCID is linked and 'Junyoung Park' is a common Korean name; this profile is name-resolved from public bibliographic data and homonyms may be merged or split.
- Both catalogued papers are 2026 arXiv preprints; peer-review status is not reflected in the input data.
- Output, impact, and activity dimensions have insufficient underlying data and are reported as missing, not as zero.