Graph ML researcher at UESTC (h-index 45) applying optimal transport to graph prompting and graph foundation models
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
45
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-09) 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 scored: this axis would measure total publication volume and its distribution across years and venues. The current snapshot exposes recent3yWorks (54) and yearly counts, but not the verified aggregate totals the axis requires, so no score is estimated.
Impact—
Not yet scored: this axis would measure citation-based influence, such as field-normalized citations or highly cited papers. Although the snapshot lists h-index 45 and i10-index 80, the full citation record needed for scoring is unavailable, so no score is estimated.
Focus0/100
The radar records a focus score of 0. The underlying field distribution classifies all matched works under one field — machine learning (graph representation learning) — at 0.5 under each of two label spellings. Topic counts add texture: advanced clustering algorithms (90) and face and expression recognition (79) lead, ahead of graph neural networks (61), complex network analysis (22), and graph theory and algorithms (22).
Collaboration36/100
Scores 36, consistent with a compact network: 54 works over the recent three years involve only 5 distinct coauthors, and both 2026 catalog papers are last-author, non-corresponding contributions. A recurring five-person pool — Xiangyu Wang, Shuo Wang, Ruiyi Fang, Xiaodong He, and Xincheng Wang — suggests a stable small team rather than broad external collaboration.
Activity100/100
Scores 100. recent3yWorks is 54, and countsByYear shows near-uninterrupted yearly output since 2013 (1 work), climbing through 2016–2017 (9 and 14), peaking at 24 works in 2019, and staying high at 16 (2020), 17 (2021), 23 (2024), 16 (2025), and 15 (2026).
Reproducibility100/100
Scores 100, with reproMean 100 across 2 reviewed papers; both 2026 arXiv catalog papers carry high repro-readiness ratings. Evolve's verdicts reinforce this: MINT is judged to have exact theory and broad evaluation, and SCGFM-ART sound theory with large reported speedups (44–85×). The evidence base is small, so read the score as an early, favorable signal.
Research summary
Zhao Kang is a researcher at the University of Electronic Science and Technology of China whose OpenAlex-indexed record spans 2013–2026, with an h-index of 45 and an i10-index of 80. Classified works concentrate in machine learning (graph representation learning), while topic counts span advanced clustering algorithms (90), face and expression recognition (79), graph neural networks (61), complex network analysis (22), and graph theory and algorithms (22). Activity is the strongest radar dimension (100): 54 works in the last three years and near-continuous yearly output, peaking at 24 works in 2019 and holding at 23 (2024), 16 (2025), and 15 (2026). Recent research centers on optimal transport for graph learning. Two 2026 arXiv papers sit in the Evolve catalog — MINT, which casts graph prompting as fixed-marginal optimal transport, and SCGFM-ART, which builds structure-centric graph foundation models through amortized Gromov-Wasserstein transport — in both cases with last-author roles. Evolve's verdicts credit exact, sound theory and large reported speedups (44–85×) while flagging small empirical deltas: roughly 1-point gains without significance tests, and near-zero accuracy effects in MINT's own controls. Both papers rate high on repro-readiness (reproMean 100 across 2 papers), and collaboration is compact, with 5 distinct coauthors over the recent three years.
Collaboration pattern
Collaboration is compact: the recent three years show 54 works with only 5 distinct coauthors (radar collaboration score 36), and both 2026 catalog papers list Zhao Kang as last author without corresponding-author status. The recurring pool — Xiangyu Wang, Shuo Wang, Ruiyi Fang, Xiaodong He, and Xincheng Wang — points to a stable small team and a supervisory last-author pattern rather than wide external networks.
Research keywords
Highlights
Amortized Gromov-Wasserstein transport for graph foundation models (SCGFM-ART)
Measure-constrained optimal transport for graph prompting (MINT)
High reproducibility readiness across reviewed work
Evolve's two reviewed 2026 papers both carry high repro-readiness ratings, driving a reproducibility radar score of 100 (reproMean 100 over 2 papers) — a consistent but still early signal, since only two papers have been assessed.
Timeline
- 2013First year in the OpenAlex record, with 1 publication
- 2017Annual output climbs to 14 works, up from 9 in 2016
- 2019Peak annual output of 24 works
- 2024Renewed high activity with 23 works
- 202615 works recorded; two optimal-transport papers (MINT on graph prompting; SCGFM-ART on graph foundation models) enter the Evolve catalog as last-author contributions
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.
- Profile is name-resolved from public bibliographic data (OpenAlex identity record, ORCID 0000-0003-4103-0954); no biographical or award information is included.
- h-index, i10-index, yearly counts, and topic counts come from the provided OpenAlex snapshot; no independent cross-database verification was performed.
- Output and impact dimensions are unscored because the underlying aggregate data was not provided (dimMissing).
- The reproducibility score (100) is computed from only 2 reviewed papers, so treat it as an early signal.