NLP researcher at Seoul National University working on computational argumentation and LLM agent reliability
3
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
1
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-17) 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—
Output metrics are not yet available: the recorded h-index (1) and i10-index (0) are too sparse to characterize total production, so this axis is left unscored pending fuller citation coverage.
Impact—
Impact cannot yet be scored: citation counts and i10-index (0) are absent or near-zero in the current OpenAlex record, so this axis awaits data rather than receiving an estimate.
Focus0/100
The focus score is 0/100, reflecting low topical concentration: the record splits evenly (0.5/0.5) between computational argumentation and LLM agent reliability, and OpenAlex topic assignments span five clusters — topic modeling (20), speech and dialogue systems (16), NLP techniques (13), multimodal ML applications (8), and information retrieval and search behavior (7). Such breadth looks typical of a record still consolidating its direction.
Collaboration54/100
Collaboration scores 54/100: the record lists 14 distinct coauthors over the last three years, and Evolve identifies 12 frequent collaborators — including Eunbeen Son, Joonsuk Park, JinYeong Bak, and Joonseok Lee — suggesting a moderately sized, actively maintained co-author network typical of group-based NLP research rather than large-scale consortia.
Activity100/100
Activity is the standout axis at 100/100: 44 works are recorded in the last three years, with yearly output climbing from 3 (2024) and 4 (2025) to 37 (2026) — an inflection consistent with the three 2026 preprints Evolve reviewed. Such a sudden jump may also reflect indexing or affiliation changes and should be read with that caution.
Reproducibility55/100
Reproducibility scores 55/100 (mean across 2 assessed papers), and all three Evolve-reviewed papers carry a medium readiness rating. Reviewer notes flag concrete gaps: no consistent mitigation for agent overtrust, no significance testing or release statement for the hallucination benchmark, and evaluation rubrics that favor personas in counterargument generation.
Research summary
Yohan Jo (Seoul National University, South Korea) is an emerging NLP researcher whose recorded work centers on two themes: computational argumentation and the reliability of LLM-based agents. The bibliometric record attributes 44 works to the last three years, with output climbing from 3 works in 2024 and 4 in 2025 to 37 in 2026, yielding a maximal activity score (100/100). The topical profile is broad: topic modeling (20 works), speech and dialogue systems (16), NLP techniques (13), multimodal machine learning applications (8), and information retrieval and search behavior (7); the even 0.5/0.5 split between computational argumentation and LLM agent reliability explains the low focus score (0/100). Evolve has reviewed three 2026 papers: persona-guided tree-based counterargument generation; a study finding that 14 LLM agents adopt corrupted tool returns (68% for search) while hiding conflicts they detect; and a multi-image object-hallucination benchmark spanning 4 tasks and 29 models. Collaboration is moderate (54/100) with 14 distinct coauthors, including Eunbeen Son, Joonsuk Park, and JinYeong Bak. Reproducibility stands at 55/100 across two assessed papers, with all three reviewed papers rated medium readiness. Citation-based output and impact scores are not yet computed (h-index 1 and i10-index 0 as recorded), so those axes remain unscored in this profile.
Collaboration pattern
The pattern is group-centered and moderately collaborative: 14 distinct coauthors in three years (score 54/100) and 12 frequent collaborators, notably Eunbeen Son, Joonsuk Park, JinYeong Bak, Joonki Min, and Joonseok Lee. On reviewed papers Jo moves between middle and last author — senior (last) position on the agent-overtrust study — and no reviewed paper is marked corresponding, suggesting shared leadership across argumentation and agent-evaluation projects.
Research keywords
Highlights
Persona-guided tree search yields more persuasive counterarguments
Measuring how much LLM agents overtrust their tools
A fine-grained benchmark for multi-image object hallucination
Timeline
- 2024First year in the bibliometric record: 3 works attributed.
- 2025Output remains small at 4 recorded works, preceding the 2026 inflection.
- 202637 works recorded; Evolve reviews three 2026 preprints — persona-guided counterargument generation, agent tool-overtrust measurement, and a multi-image hallucination benchmark.
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.
- All bibliometrics come from a single OpenAlex record; no ORCID is linked, so homonymous attribution cannot be ruled out.
- Output and impact axes are unscored: the recorded h-index (1) and i10-index (0) are too sparse to characterize citations.
- All three reviewed papers are 2026 arXiv preprints; peer-reviewed venue status is not reflected in this snapshot.
- The jump from 4 works (2025) to 37 (2026) may partly reflect indexing or affiliation changes rather than pure productivity.