University of Auckland researcher building Chinese argumentation datasets and AI benchmark evaluation infrastructure
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—
Output is unscored because no publication-count or venue data is available yet beyond the two 2026 catalog entries. This axis would measure total research volume — papers per year, venue types and authorship roles — across the full publication record.
Impact—
Impact is unscored because no citation or influence metrics are available in the current data. This axis would measure citations, downstream reuse of the released datasets and search tools, and uptake of the benchmarks by the evaluation community.
Focus0/100
The focus score is 0 because the two cataloged papers split the field distribution evenly — 0.5 in natural language processing (computational argumentation) and 0.5 in AI evaluation infrastructure / information retrieval. The portfolio is therefore a deliberate two-track profile rather than a single dominant line: Chinese-language debate benchmarking on one side, benchmark search and catalog infrastructure on the other, both converging on LLM evaluation.
Collaboration56/100
The collaboration score is 56, based on 16 distinct coauthors recorded in the bibliometric data. Both Evolve-reviewed papers are middle-authorship, non-corresponding team contributions rather than solo or lead-author work. A recurring collaborator pool — including Zongrui Yang, Haoyuan Li, Yi Zhou and Jiamou Liu — indicates a stable, moderately sized network supporting the two 2026 projects.
Activity—
Activity is unscored: no year-by-year publication cadence is available beyond the two 2026 catalog entries. This axis would measure recent output rate and the consistency of publishing over time, which cannot yet be established from a single catalog year.
Reproducibility78/100
Reproducibility is the strongest scored axis at 78, the mean readiness across the two reviewed papers. Benchmark Radar rates high, supported by strong provenance over 1,283 benchmark records, while the Chinese debating benchmark rates medium: its design and label-reliability analysis are solid but its results lack significance tests. The pattern points to deliberate attention to data provenance and reuse.
Research summary
Pengqian Han is a researcher at the University of Auckland whose Evolve-visible work sits at the intersection of computational argumentation and AI evaluation infrastructure. The catalog contains two 2026 arXiv preprints, both middle-authorship, non-corresponding contributions. The first, Chinese Competitive Debating Dataset and Benchmark, provides an expert-annotated Chinese debate resource spanning 148 matches and 3 tasks; Evolve's review judges the design and label-reliability analysis solid while noting the results lack significance tests. The second, Benchmark Radar, builds a living database and search engine over 1,283 AI benchmark records, distinguished by a rare full-catalog census and strong provenance, though its own retrieval quality remains unevaluated. Together they sketch a two-track profile — one track creating language-specific argumentation benchmarks, the other building retrieval and cataloging infrastructure the evaluation ecosystem lacks — unified by an emphasis on LLM evaluation. Bibliometric data records 16 distinct coauthors (collaboration score 56) and an evenly split field distribution across the two tracks (focus score 0). Reproducibility is the strongest signal, with a mean readiness of 78 across the two reviewed papers; Benchmark Radar rates high and the debating benchmark medium. Output, impact and activity axes are unscored because the underlying bibliometric data is insufficient, so this should be read as an early-stage, catalog-based profile.
Collaboration pattern
The collaboration score of 56 reflects 16 distinct coauthors across the bibliometric record. Both cataloged papers are middle-authorship team efforts rather than solo or corresponding-author work. The frequent-coauthor list — including Zongrui Yang, Haoyuan Li, Zhongsheng Wang, Zhirui Zeng, Yi Zhou (listed twice, likely a recurring partner), Yuting Wang, Jiamou Liu, Koutian Wu, Junjie Zhou, Ergan Shang and Jiayu Wang — points to a stable, moderately sized collaboration cluster behind the two 2026 projects.
Research keywords
Highlights
Expert-annotated Chinese competitive debating benchmark
Benchmark Radar: a living search engine for AI benchmarks
Reproducibility-forward evaluation tooling
Across the two Evolve-reviewed papers the mean reproducibility readiness is 78: Benchmark Radar is rated high on documented provenance, while the Chinese debating benchmark is rated medium. The consistent pattern is reusable data and documented sources aimed at LLM evaluation.
Timeline
- 2026Co-authors an expert-annotated Chinese competitive debating dataset and benchmark on arXiv, spanning 148 matches and 3 tasks.
- 2026Co-authors Benchmark Radar on arXiv, a living database and search engine covering 1,283 AI benchmark records with a full-catalog census.
- 2026Year-end catalog state: two preprints spanning computational argumentation and AI evaluation infrastructure, 16 distinct coauthors, mean reproducibility readiness 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.
- Bibliometric signals derive from OpenAlex-sourced public data; output, impact and activity axes lack sufficient underlying data and are left unscored.
- This profile is name-resolved from public bibliographic data (ORCID 0000-0002-6684-9010); identity homonyms cannot be fully ruled out.
- Qualitative assessments rest on only two Evolve-reviewed 2026 arXiv preprints; peer-reviewed venue status is unknown.
- The frequent-coauthor list appears to contain a duplicate name variant (Yi Zhou listed twice), so the count of distinct recurring collaborators may differ.