Early-career researcher at Shanghai Jiao Tong University in LLM multi-agent systems, document AI, and robot learning
3
papers here
3
reviewed by Evolve
2
first-author
1
corresponding
—
works · OpenAlex
1
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 yet: the data available lacks a complete, verified publication record beyond OpenAlex aggregates for this author. This axis would measure total output volume, venue distribution, and growth trajectory across the full career once a fuller record exists.
Impact—
Impact cannot be scored yet: citation-linked evidence is too thin to estimate stably (h-index 1, i10-index 0). This axis would measure citations, citing works, and influence on follow-up research once the citation record matures beyond this early baseline.
Focus0/100
The focus score of 0 reflects breadth rather than single-line depth: the reviewed work splits evenly, about 0.33 each, across vision-language-action robot learning, LLM multi-agent systems, and document AI with citation attribution. OpenAlex topic counts reinforce the spread — multimodal machine learning applications (16), topic modeling (13), NLP techniques (7), reinforcement learning in robotics (7), and domain adaptation / few-shot learning (7) — indicating a deliberately multi-field early agenda.
Collaboration54/100
The collaboration score of 54 sits mid-range: OpenAlex records 14 distinct coauthors against 40 recent works, a moderate pool for that volume. Twelve recurring co-authors appear in the Evolve network, including Chenyu Su, Zhaolong Shen, Yuan Qian, Rui Zhang, and Feng Yan. Across the three catalog papers Chen Qian holds two first-author roles and one middle authorship, consistent with small-team leadership rather than large-consortium work.
Activity100/100
The activity score of 100 is the strongest signal in the record: 40 works within the recent three-year window, with OpenAlex year counts jumping from 1 in 2025 to 39 in 2026. All three Evolve-reviewed papers are 2026 arXiv outputs and two are first-authored, indicating an intense, current, largely self-driven production tempo.
Reproducibility43/100
The reproducibility score of 43 is the mean over 3 Evolve-assessed papers. Readiness is medium for the multi-agent debate analysis and for AtomCite, and low for VLA-Precision, where Evolve flagged confounded baselines and thin statistics despite strong hardware evidence. The pattern suggests benchmark-first work is better scaffolded for replication than the real-robot RL claims.
Research summary
Chen Qian is an early-career researcher at Shanghai Jiao Tong University whose name-resolved OpenAlex record opens in 2025 with one indexed work and jumps to 39 works in 2026 — 40 works across the recent three-year window. The record splits almost evenly (0.33 each) across three fields: vision-language-action robot learning, LLM multi-agent systems, and document AI with citation attribution. OpenAlex topic clustering mirrors that breadth, led by multimodal machine learning applications (16), topic modeling (13), natural language processing techniques (7), reinforcement learning in robotics (7), and domain adaptation / few-shot learning (7). Within the Evolve-reviewed catalog, Chen Qian is first author on two of three papers — a layered analysis of multi-agent LLM debate (also corresponding author) and AtomCite, a benchmark and framework for verifying and correcting page-level citations — and a middle author on VLA-Precision, which applies real-world online RL to vision-language-action models. The citation footprint is nascent (h-index 1, i10-index 0), so output and impact radar axes are left unscored. Evolve reproducibility assessments average 43 across the three papers (medium, medium, low readiness). The collaboration pool holds 14 distinct coauthors, with recurring names including Chenyu Su, Zhaolong Shen, and Yuan Qian. Read as a whole, this is a high-velocity, multi-line early-stage profile whose citation evidence has not yet caught up with its output.
Collaboration pattern
Collaboration is moderate-scale and team-based: 14 distinct coauthors across 40 recent works, with twelve recurring names led by Chenyu Su, Zhaolong Shen, and Yuan Qian. Across the three Evolve-reviewed papers Chen Qian takes first authorship twice (corresponding on the debate analysis) and a middle position on the VLA robotics paper, consistent with leading small NLP teams while contributing as co-author in robotics. No large multi-institution consortium pattern is visible in the current record.
Research keywords
Highlights
First-author analysis: LLM debate shifts tone, not answer quality
AtomCite: first benchmark for page-level citation verification
VLA-Precision reports 98.3% success via real-world online RL
Timeline
- 2025The name-resolved OpenAlex record opens with 1 indexed work in the year counts, marking the start of the bibliometric trail.
- 2026Output surge: 39 works are recorded in 2026; within this year the layered analysis of multi-agent LLM debate appears on arXiv with Chen Qian as first and corresponding author.
- 2026AtomCite, the first benchmark and framework for page-level citation verification and correction in multi-page documents, is released on arXiv (first author).
- 2026VLA-Precision on efficient real-world online RL for vision-language-action models appears on arXiv (middle author).
Papers in the Evolve catalog
- first authorcorresponding
What Does Multi-Agent LLM Debate Actually Change? A Layered Analysis of Disagreement and Answer Quality
- first author
AtomCite: Verification and Correction of Supplied Page-Level Citations in Multi-Page Documents
VLA-Precision: Asymmetric Co-Bootstrapping for Efficient Real-World Online RL of Vision-Language-Action Models
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, so name disambiguation relies on public bibliographic data — 'Chen Qian' is a common name.
- Citation-impact evidence is effectively absent (h-index 1, i10-index 0); the output and impact radar axes are left unscored rather than estimated.
- Only 3 papers carry Evolve review coverage, against 40 recent works in the bibliometric record; the qualitative verdicts sample a small slice of total output.
- The reproducibility mean (43) rests on just 3 assessed papers, so it is sensitive to any single paper's readiness rating.