Graph-ML researcher at South China Normal University applying discrete curvature to GNN sampling and pooling.
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
0
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-15) 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 measures publication volume and its growth over time. It is unscored here: the OpenAlex snapshot records h-index and i10-index as 0 with all indexed activity in 2026, so the record is too new to rate.
Impact—
Impact tracks citations and downstream influence on follow-up research. It is unscored: all indexed works date to 2026 and the recorded h-index and i10-index are 0, so citation evidence has not accumulated enough to support a rating.
Focus0/100
Focus is scored 0/100, although the raw footprint looks narrow: fieldDist splits 0.5/0.5 between 'graph machine learning' and its geometric-deep-learning label — two near-identical tags — while topic counts concentrate on one discrete-curvature thread, led by Advanced Graph Neural Networks (6), Topological and Geometric Data Analysis (5), and Geometric Analysis and Curvature Flows (4), with Graph Theory and Algorithms (2) and stochastic gradient optimization (1) as satellites.
Collaboration39/100
Collaboration scores 39/100, with 6 distinct coauthors in the snapshot. Recurring partners include Chaoqun Fei, Guoxuan Li, Chuanqing Wang, Yangyang Li, and Tianyong Hao. Both reviewed papers place Zhou in a middle author position without corresponding-author status, consistent with contributing inside a small, stable team rather than leading a large or international consortium.
Activity57/100
Activity scores 57/100. countsByYear records 6 works in 2026 — the only indexed year — and recent3yWorks is also 6, so the entire visible OpenAlex footprint is a single-year burst of arXiv preprints rather than a sustained multi-year cadence; the score reflects that density and recency, with no earlier years to carry it.
Reproducibility55/100
Reproducibility averages 55/100 across the 2 scored papers, both rated medium readiness. For ERC-LG, Evolve credits credible scaling results but finds accuracy wins mostly within noise; for RicciPool, ablations are thorough, yet reported wins lack significance tests and some numbers conflict — practical to re-run, but statistical claims need care.
Research summary
Tinglve Zhou is an early-stage researcher at South China Normal University (China) working in graph machine learning, with the record split evenly between graph machine learning and its geometric deep learning branch. The OpenAlex footprint is entirely recent: countsByYear shows 6 works in 2026 — the only indexed year — and recent3yWorks confirms the same 6, so this is a name-resolved, preprint-driven profile built on arXiv postings rather than an established citation record (h-index and i10-index are recorded as 0). Evolve has reviewed two of these papers, both with Zhou in a middle author position and non-corresponding. 'Scalable Subgraph Sampling via Resistance Curvature' (2026) proposes ERC-LG, combining Johnson-Lindenstrauss projection with batched conjugate-gradient solves for curvature-guided GNN subgraph sampling; Evolve judges the scaling results credible while accuracy gains mostly fall within noise. 'Geometric Flow enhanced Graph Coarsening' (2026) contributes RicciPool, adding Ollivier-Ricci flow edge re-weighting to graph pooling with thorough ablations, though reported wins lack significance tests and some numbers conflict. Topic counts reinforce a coherent discrete-curvature agenda: Advanced Graph Neural Networks (6), Topological and Geometric Data Analysis (5), Geometric Analysis and Curvature Flows (4), Graph Theory and Algorithms (2), and stochastic gradient optimization (1). Reproducibility averages 55/100 across the two reviewed papers, both rated medium readiness, while collaboration scores 39/100 with 6 distinct coauthors. Overall, this is a tightly focused geometric graph-ML agenda in its first visible year.
Collaboration pattern
The snapshot lists 6 distinct coauthors, with a recurring core of Chaoqun Fei, Guoxuan Li, Chuanqing Wang, Yangyang Li, and Tianyong Hao — a small, stable team pattern. On both reviewed papers Zhou holds a middle author position without corresponding-author status, consistent with contributing to group projects rather than directing them; the collaboration dimension scores 39/100, reflecting limited demonstrated breadth so far.
Research keywords
Highlights
Scalable curvature-guided subgraph sampling (ERC-LG)
Ricci-flow re-weighting for graph coarsening (RicciPool)
A coherent discrete-curvature agenda in graph ML
Both reviewed papers route discrete curvature — effective resistance on one side, Ollivier-Ricci flow on the other — into graph-ML pipelines, matching topic counts led by Advanced Graph Neural Networks (6) and Topological and Geometric Data Analysis (5).
Timeline
- 2026OpenAlex record begins: 6 works indexed in 2026, the only year with visible activity (h-index and i10-index recorded as 0).
- 2026arXiv preprint 'Scalable Subgraph Sampling via Resistance Curvature' proposes ERC-LG, using Johnson-Lindenstrauss projection and batched conjugate gradient for curvature-guided GNN subgraph sampling.
- 2026arXiv preprint 'Geometric Flow enhanced Graph Coarsening' proposes RicciPool, adding Ollivier-Ricci flow edge re-weighting to graph pooling.
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.
- Bibliometrics are computed from an OpenAlex snapshot only (identity: openalex); no ORCID was available, so this profile is name-resolved from public bibliographic data.
- Output and impact dimensions are unscored: all indexed works fall in 2026 and h-index/i10-index are recorded as 0, so the citation record has not matured.
- The reproducibility mean (55/100) rests on only 2 reviewed papers, so it should be read as preliminary.
- Evolve's qualitative verdicts cover 2 of the 6 indexed 2026 works; the remaining works are not yet assessed.