Beijing Institute of Technology researcher on LLM multi-agent systems, agent memory, and multimodal graph learning.
3
papers here
3
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—
Not yet scorable: this profile draws on only three 2026 arXiv preprints, with no full publication count, venue history, or authorship-share data. A complete output axis would measure cumulative papers, publication venues, and first-author share over a full career.
Impact—
Citation-based impact cannot be scored: the supplied data contain no citation counts, h-index, or attention signals. This axis would measure citations to the three preprints and the uptake of the MACE, OpenMAS-GCom, and BRAIN artifacts by other work.
Focus0/100
Focus scores 0 because activity is split perfectly evenly across three subfields — LLM-based multi-agent systems, their evaluation, and graph representation learning, each exactly one third of the field distribution. The two strands are technically coherent, since graphs underpin both, but the portfolio is broad rather than concentrated in a single niche.
Collaboration60/100
Collaboration scores 60 on the strength of 19 distinct co-authors across just three papers, indicating genuinely team-based research. Recurring names — Xunkai Li, Rong-Hua Li, Kairui Yang, Minghao An — anchor a stable Beijing Institute of Technology cluster, while Chen occupies middle author positions on every catalogued paper rather than a leading slot.
Activity—
Year-by-year activity cannot be scored: every catalogued paper carries 2026, so no multi-year publication cadence, continuity, or latency trend can be computed. This axis would track papers per year and research consistency over time as the record accumulates.
Reproducibility55/100
Reproducibility averages 55 across 3 assessed papers, each rated medium readiness: MACE, OpenMAS-GCom, and BRAIN describe methods and benchmarks in enough detail for replication attempts, but Evolve's reviews flag single-run evaluations, sub-0.1pp margins, and untested transfer that replication would need to resolve.
Research summary
Zekai Chen is a researcher at Beijing Institute of Technology whose recent work, captured in three 2026 arXiv preprints in the Evolve catalog, sits at the intersection of LLM-based multi-agent systems and graph representation learning. All three papers list Chen in a middle author position with no corresponding-author role, indicating substantive contribution within larger teams anchored by frequent co-authors such as Xunkai Li and Rong-Hua Li. The first paper, MACE, proposes memory-agent co-evolution with adaptive memory graphs, reporting 81.11% versus SAGE's 78.97% across 8 benchmarks, though Evolve's review notes single-run evaluations, sub-0.1pp margins, and uncited baselines. The second, OpenMAS-GCom, is a controlled-intervention diagnostic benchmark for graph-enhanced multi-agent systems spanning 29 datasets, with intervention evidence on five of them. The third, BRAIN, is a bridge-router-adapter approach toward unified multimodal graph foundation models, evaluated on 9 datasets with strong but narrow margins. Bibliometric indicators show 19 distinct co-authors, an even one-third split across three subfields, a collaboration score of 60, and a mean reproducibility readiness of 55 across 3 assessed papers, all rated medium. Citation impact, total output, and year-by-year activity metrics are not yet available, so this should be read as an early-stage, name-resolved profile built from public bibliographic data.
Collaboration pattern
With 19 distinct co-authors across three 2026 papers, Chen works in sizeable, stable teams. Xunkai Li and Rong-Hua Li recur most often and likely anchor the Beijing Institute of Technology group, with Kairui Yang, Minghao An, Kaixiang Zhang, Yuxuan Ba, Ziheng Yi, and Guangyuan He rounding out the pool. Chen holds middle author positions on all catalogued papers, with no corresponding-author role yet — a core-contributor profile rather than a team lead.
Research keywords
Highlights
MACE pushes memory-agent co-evolution to 81.11% across 8 benchmarks
A diagnostic benchmark dissecting graph-enhanced multi-agent systems
BRAIN routes modality choice for multimodal graph foundation models
Timeline
- 2026Publishes MACE, proposing memory-agent co-evolution with adaptive memory graphs for multi-agent systems (arXiv, middle author).
- 2026Publishes OpenMAS-GCom, a controlled-intervention diagnostic benchmark for graph-enhanced multi-agent systems (arXiv, middle author).
- 2026Publishes the BRAIN bridge-router-adapter approach toward a unified multimodal graph foundation model (arXiv, middle author).
- 2026All three catalogued outputs appear within a single year as team-based arXiv preprints, marking a productive early-stage record.
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 indicators are computed from a public bibliographic snapshot; citation counts, h-index, and venue-level data are not included.
- All catalogued outputs are 2026 arXiv preprints and may not yet be peer-reviewed.
- No ORCID is linked; 'Zekai Chen' is a common name resolved from public bibliographic data, so author attribution carries some ambiguity.
- Output, impact, and activity axes are unscored because the underlying longitudinal and citation data are unavailable.