LLM benchmarking researcher: AI exploration in verifiable worlds and fine-grained novelty assessment for peer review.
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
0
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—
Output cannot be scored: the snapshot provides no productivity series beyond 3 works dated 2026, so publication volume is insufficient. This axis measures total works, publication cadence over time, and authorship roles.
Impact—
Citation impact cannot be scored yet: the snapshot reports h-index 0 and i10-index 0 with works dated only 2026, so none has had time to accrue citations. This axis would measure citations, h-index, and downstream influence.
Focus0/100
The focus score is 0, reflecting topical spread in the OpenAlex topics: five topics each appear once — topic modeling, scientometrics, robotic reinforcement learning, AI planning, and forest insect ecology — a pattern that likely includes same-name resolution noise. Within the reviewed catalog, however, focus is coherent: fieldDist splits 0.5/0.5 between LLM agent evaluation and LLM evaluation for peer review.
Collaboration70/100
Collaboration scores 70 on the strength of a wide co-author network: 33 distinct coauthors appear in the recent three-year window against only 3 recorded works, and 12 recur frequently, including Ming Zhang, Shihan Dou, and Junjie Ye. Both catalog papers are middle-authored and non-corresponding, indicating contribution within larger team-led benchmark projects rather than sole leadership.
Activity40/100
Activity scores 40: OpenAlex logs 3 works in the recent three-year window, and countsByYear shows all 3 in 2026 with nothing recorded for earlier years. Both Evolve-reviewed papers appeared on arXiv in 2026, so recorded output is recent and front-loaded; a longer-term cadence cannot yet be established.
Reproducibility60/100
Reproducibility scores 60 (reproMean over 2 reviewed papers). NovGauge rates high — evidence-grounded construction supports replication — while ExplorationBench rates low, with n=3 trajectories and single-run controls limiting statistical strength and reuse. The mean signals sound benchmark methodology overall, with clear per-paper variance.
Research summary
Zhonghan Yue (Xinjiang Normal University, China) is an early-stage researcher whose OpenAlex-resolved profile centers on benchmarking and evaluating large language models. The Evolve catalog reviews two 2026 arXiv papers with Yue as a middle author: ExplorationBench, which measures AI systems' exploration in verifiable alien worlds with exact grading, and NovGauge, a fine-grained benchmark for diagnosing LLMs' capability in paper novelty assessment. The field distribution splits evenly (0.5/0.5) between LLM agent evaluation and LLM evaluation for peer review, so the visible work forms one coherent agenda: building verifiable benchmarks that stress-test LLM and agent capabilities. Collaboration is the strongest measured axis (70): 33 distinct coauthors appear in the recent three-year window against only 3 recorded works, with 12 recurring collaborators including Ming Zhang, Shihan Dou, and Junjie Ye — a pattern consistent with contributing defined modules inside larger team-led projects. Reproducibility scores 60 across the two reviewed papers: high for NovGauge, low for ExplorationBench, whose n=3 trajectories and single-run controls limit statistical strength. Activity scores 40, with all recorded output dated 2026. Citation impact cannot yet be assessed: the snapshot reports h-index 0 and i10-index 0, and no works predate 2026. The broader topic spread (five single-count topics, including forest insect ecology) likely reflects same-name resolution noise and is flagged in the caveats.
Collaboration pattern
Collaboration is the strongest measured axis (70/100): the recent three-year window shows 33 distinct coauthors against only 3 recorded works, a wide, team-embedded pattern. Twelve collaborators recur, including Ming Zhang, Zhenghao Xiang, Shihan Dou, and Junjie Ye. In both reviewed papers Yue is a middle author and not corresponding, consistent with contributing defined modules such as benchmark design and evaluation pipelines inside larger LLM-evaluation group projects.
Research keywords
Highlights
ExplorationBench: exact-grading benchmarks for AI exploration
NovGauge: fine-grained diagnosis of LLM novelty assessment
Timeline
- 2026ExplorationBench posted to arXiv: a benchmark measuring AI systems' exploration in verifiable alien worlds (middle author).
- 2026NovGauge posted to arXiv: a fine-grained benchmark for diagnosing LLMs' paper novelty assessment capability (middle author).
- 2026OpenAlex logs 3 works in the recent three-year window, all dated 2026 — the profile's entire recorded output.
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 are OpenAlex-derived with no ORCID link; identity is name-resolved from public bibliographic data, and off-field topics (e.g., forest insect ecology) likely reflect same-name entity noise.
- Citation impact is unassessable for now: h-index and i10-index are 0 and all indexed works date to 2026.
- Only two papers, both arXiv preprints with middle-author, non-corresponding roles, are reviewed in the Evolve catalog.
- countsByYear covers only 2026 (3 works); no longer-term productivity trend is available.