Zihan Mei
University of Electronic Science and Technology of China
Graph ML researcher at UESTC: signed-graph pre-training & prompt learning; plastic-memory graph few-shot CIL
2
papers here
2
reviewed by Evolve
2
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-08) 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 scored: the snapshot lacks verified publication counts, venue tiers, and output volume for this name, so this axis cannot be computed. Once such records are linked, it would measure the quantity, venue quality, and consistency of the researcher's publication record.
Impact—
Not scored: no citation counts, h-index, or downstream-usage signals are available in the data, and both catalog entries are recent 2026 arXiv preprints. This axis would measure citation-based influence and how far other researchers adopt or build on the work.
Focus0/100
Scores 0 because the field distribution splits evenly at 0.5/0.5 between two labels — 'graph machine learning (signed graph representation learning / topological data analysis)' and 'machine learning (graph learning)' — with no single dominant field. Substantively, both halves sit inside graph machine learning, so the low score reflects labeling granularity rather than genuinely scattered research interests.
Collaboration46/100
Scores 46. The snapshot records 9 distinct co-authors across the reviewed work, including Rong Pan, Yuzhou Chen, Yixuan He, Zhili Qin, Tongze Zhang, Hongyuan Liu, Junming Shao, and Qinli Yang. The network is moderate in size and concentrated in a stable recurring team surrounding the two 2026 first-author preprints, rather than a broad cross-institution web of partners.
Activity—
Not scored: the data provides no yearly publication counts or cadence over time — only two 2026 catalog entries. This axis would measure research pace and continuity, such as papers per year, recency of output, and sustained engagement across consecutive periods.
Reproducibility78/100
Scores 78, computed as the mean reproducibility rating across 2 reviewed papers. The signed-graph pre-training paper (TopoSIGN) rates high on reproducibility readiness, with experiments judged transparent, while the plastic-memory framework for graph few-shot class-incremental learning rates medium — Evolve notes missing variance statistics and a thin margin on CoraFull.
Research summary
Zihan Mei is a graph machine learning researcher at the University of Electronic Science and Technology of China whose Evolve catalog to date consists of two first-authored 2026 arXiv preprints, each opening a distinct thread within graph learning. The first, Signed Graph Pre-Training and Prompt Learning (TopoSIGN), is described by Evolve as the first signed-graph pre-training and prompt framework, combining a signed Dowker-filtration topology branch with persistent homology and a magnetic signed Laplacian; its experiments are judged transparent, though semi-supervised variants outperform the full pipeline and topology gains remain inconsistent. The second, A Lightweight Plastic-Memory Framework for Graph Few-Shot Class-Incremental Learning, tackles catastrophic forgetting with micro-cluster plastic memory and dual-loop meta-learning built on prototypical networks, reporting strong efficiency and accuracy gains alongside thinner evidence — no variance statistics and a narrow margin on CoraFull. Bibliometrically, the snapshot records 9 distinct co-authors (collaboration 46), an even 0.5/0.5 field split between signed-graph/topological methods and graph learning (focus 0), and a reproducibility mean of 78 across 2 papers, ranging from high (TopoSIGN) to medium readiness. Output, impact, and activity axes remain unscored for lack of data, so this reads as an early-stage, name-resolved profile rather than a complete record.
Collaboration pattern
The snapshot counts 9 distinct co-authors (collaboration score 46) with a stable recurring team: Rong Pan, Yuzhou Chen, Yixuan He, Zhili Qin, Tongze Zhang, Hongyuan Liu, Junming Shao, and Qinli Yang. Both 2026 first-author preprints sit inside this circle, suggesting close-knit in-group collaboration at UESTC rather than broad external partnerships; the moderate score reflects a small, concentrated network at an early career stage.
Research keywords
Highlights
TopoSIGN: first signed-graph pre-training and prompt framework
Plastic memory tackles forgetting in graph few-shot CIL
Reproducibility mean of 78 across two reviewed papers
Evolve rates reproducibility readiness high for TopoSIGN and medium for the plastic-memory framework, averaging 78 across 2 papers — transparent experiment reporting on one side, unreported variance statistics on the other.
Timeline
- 2026Posts first-author arXiv preprint 'Signed Graph Pre-Training and Prompt Learning' (TopoSIGN), positioned as the first signed-graph pre-training and prompt framework.
- 2026Posts first-author arXiv preprint 'A Lightweight Plastic-Memory Framework for Graph Few-Shot Class-Incremental Learning', applying micro-cluster plastic memory and dual-loop meta-learning to catastrophic forgetting.
- 2026Evolve reviews both preprints, rating reproducibility readiness high for TopoSIGN and medium for the plastic-memory framework, with a reproducibility mean of 78 across the 2 papers.
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 scores (collaboration 46, focus 0, reproducibility 78) come only from the provided bibliometric snapshot; output, impact, and activity axes lack sufficient underlying data to score.
- Both catalog entries are 2026 arXiv preprints; peer-review status, venue acceptance, and citation history are outside this dataset.
- The catalog includes only papers reviewed by Evolve; other publications by this researcher may exist but are out of scope, so output may be under-counted.
- No ORCID or country is attached and 'Zihan Mei' is a common Chinese name; the profile is name-resolved from public bibliographic data and may merge or split identities.