Rong-Hua Li

Beijing Institute of Technology

Unclaimed · AI-structured profileOpenAlex

Rong-Hua Li: graph learning and LLM multi-agent systems researcher at Beijing Institute of Technology

papers here

reviewed by Evolve

first-author

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 data is not yet available: this snapshot provides no publication counts or venue totals, so the axis cannot be scored. It would measure the volume and cadence of the researcher's published papers.

Impact—

Impact data is insufficient in this snapshot: no citation counts, h-index, or venue influence metrics were provided. This axis would measure how widely and how strongly the research community adopts and cites the work.

Focus0/100

The focus axis scores 0/100, reflecting research attention spread evenly across three distinct directions — LLM-based multi-agent systems, their evaluation, and graph representation learning — each accounting for roughly 33% of the reviewed field distribution. This even three-way split indicates breadth rather than a single deep specialization, though it is computed from only three reviewed papers, so the balance may shift as the catalog grows.

Collaboration60/100

The collaboration axis scores 60/100, based on 19 distinct co-authors observed in the snapshot. Recurring partners include Xunkai Li, Zekai Chen, and Kairui Yang, with Minghao An, Kaixiang Zhang, Yuxuan Ba, Ziheng Yi, and Guangyuan He also appearing across the reviewed work. This indicates an active but moderately sized network, consistent with a senior group-lead role on recent multi-agent and graph-learning projects.

Activity—

Activity data is unavailable: the snapshot contains no per-year publication time series, so recent output cadence cannot be scored. This axis would measure ongoing research productivity and momentum over time.

Reproducibility55/100

The reproducibility axis scores 55/100, the mean readiness across the three reviewed papers (reproMean 55, reproPapers 3), each rated medium. Evolve's verdicts note rigorous benchmark design in OpenMAS-GCom, but flag single-run evaluations and sub-0.1-percentage-point margins in MACE, and narrow margins with untested cross-domain transfer in the BRAIN graph foundation model. Overall this signals workable but not yet turnkey reproduction.

Research summary

Rong-Hua Li is a researcher at Beijing Institute of Technology whose recent work, as reviewed by Evolve, spans three evenly weighted directions: LLM-based multi-agent systems, their evaluation, and graph representation learning. In 2026, three arXiv papers appeared with Li in the last-author position. MACE introduces memory-agent co-evolution with adaptive memory graphs, reporting results across 8 benchmarks (81.11% versus 78.97% for SAGE), though Evolve notes single-run evaluations and sub-0.1-percentage-point margins leave the state-of-the-art claim unproven. OpenMAS-GCom proposes a controlled-intervention diagnostic benchmark covering 29 datasets for graph-enhanced multi-agent systems, rated rigorous in design even though intervention evidence spans only five datasets. The third paper, BRAIN, presents a bridge-router-adapter approach toward unified multimodal graph foundation models with strong results on 9 datasets, subject to narrow margins and untested cross-domain transfer. Bibliometrically, the snapshot records 19 distinct co-authors, recurring collaborators including Xunkai Li, Zekai Chen, and Kairui Yang, a collaboration score of 60, and a reproducibility score of 55 (mean over three papers, all rated medium readiness). The focus score is 0, reflecting the even three-way field split. Output, impact, and activity metrics are not available in this snapshot, so the profile should be read as an early-stage, catalog-grounded view of a senior contributor rather than a complete bibliometric record.

Collaboration pattern

The snapshot records 19 distinct co-authors and a collaboration score of 60. The most recurring partners are Xunkai Li, Zekai Chen, and Kairui Yang, with Minghao An, Kaixiang Zhang, Yuxuan Ba, Ziheng Yi, and Guangyuan He also appearing. All three reviewed papers list Li as last author, consistent with a senior supervisory role on team-driven LLM-agent and graph-learning projects.

Research keywords

LLM-based multi-agent systemsAgent memoryMemory graphsGraph-enhanced multi-agent systemsDiagnostic benchmarkingCommunication topologyGraph representation learningMultimodal graph learningGraph foundation modelsModality routingNode classification

Highlights

MACE: memory-graph co-evolution leads 8 benchmarks
Evolve's review finds MACE's memory-agent co-evolution with adaptive memory graphs reaching 81.11% versus SAGE's 78.97% across 8 benchmarks, while cautioning that single-run evaluations and sub-0.1pp margins leave the state-of-the-art claim unproven.
OpenMAS-GCom: controlled-intervention diagnostic benchmark
The paper claims a controlled-intervention benchmark diagnosing graph-enhanced multi-agent system components across 29 datasets. Evolve rates the design rigorous, but notes that intervention evidence currently spans only five datasets.
BRAIN: bridge-router-adapter for multimodal graph foundation models
A bridge-router-adapter architecture routes modality choice across graph scopes; Evolve reports strong results on 9 datasets, with narrow margins and untested cross-domain transfer flagged as open questions.

Timeline

2026Last-author arXiv paper MACE introduces memory-agent co-evolution with adaptive memory graphs for multi-agent systems.
2026Last-author arXiv paper OpenMAS-GCom proposes a controlled-intervention diagnostic benchmark for graph-enhanced multi-agent systems.
2026Last-author arXiv paper proposes a bridge-router-adapter approach toward a unified multimodal graph foundation model.

Papers in the Evolve catalog

Frequent co-authors

Xunkai Li×3 Zekai Chen×3 Kairui Yang×2 Minghao An×2 Xunkai Li×2 Zekai Chen×2 Rong-Hua Li×2 Kaixiang Zhang×1 Yuxuan Ba×1 Ziheng Yi×1 Guangyuan He×1 Kairui Yang×1

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 scores are derived from OpenAlex-based data only; no citation counts, h-index, or venue-level statistics were provided, so the output, impact, and activity axes are marked missing.
The profile is name-resolved from public bibliographic data; no ORCID is linked and 'Rong-Hua Li' is a common name, so identity confidence is limited.
Evidence rests on three 2026 arXiv papers reviewed by Evolve, all with Li as last author and none flagged as corresponding author; other publications are not reflected here.
Reproducibility (55) is a mean over only three papers, and country metadata beyond the Beijing Institute of Technology affiliation was not provided.