Early-Career LLM-Agent Privacy Researcher at Minjiang University: Leakage Frameworks and Platform Mediation
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
—
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 volume is not yet measurable: this snapshot contains no verified publication count, venue record, or yearly output series for Guolong Zheng. This axis would measure total and per-year research output across indexed venues once catalog coverage expands.
Impact—
Citation-based impact cannot be scored yet: no citation counts, h-index, or citing-work data are available in this snapshot. This axis would measure how widely the two cataloged 2026 preprints are taken up by later research.
Focus0/100
The focus score is 0, driven by an even field distribution of 0.5/0.5 between 'AI security & privacy (LLM agents)' and 'computer security (LLM-agent systems privacy)'. In substance the split is narrow: both cataloged 2026 papers target privacy in LLM-agent systems, one attacking leakage and one mediating propagation. The zero reading reflects labeling across two adjacent category names rather than demonstrated topic-hopping, so interpret it as an indexing artifact with caution.
Collaboration51/100
The collaboration score is 51, based on 12 distinct coauthors observed in the cataloged record. Recurring partners include Chen Hou (the most frequent), Tao Huang, Guosen Wu, Jiayang Meng, Xu Yang, Xuechao Yang, and Feng Xia. On both 2026 papers Zheng sits in middle and last author positions without corresponding authorship, suggesting contribution within a team rather than sole leadership; the moderate score reflects a small but stable early-stage network.
Activity—
Research activity cannot be scored yet: this snapshot lacks per-year publication timestamps beyond the two 2026 catalog entries. This axis would measure publication cadence and engagement over time once a longer indexed history accumulates.
Reproducibility55/100
Reproducibility scores a mean of 55 across 2 reviewed papers, with both rated medium readiness. Evolve review flags concrete gaps: CIPL's cross-provider results over 5 providers lack statistical testing, and PAPC's zero-leak results are partly by construction. The mid-range mean reflects black-box evaluation protocols that others can re-run, offset by missing significance tests and metrics whose guarantee depends on the mediation design itself.
Research summary
Guolong Zheng (Minjiang University; ORCID 0009-0000-4677-1642) is the subject of an early-stage researcher profile grounded in two 2026 arXiv preprints reviewed in the Evolve catalog, both on the security and privacy of large language model (LLM) agents. The first, CIPL, formalizes black-box agent privacy leakage as an exposure-versus-recovery problem covering RAG privacy, tool-use security, and memory extraction; its results hold coherently across 5 providers, though they lack statistical testing. The second, PAPC, introduces event-level platform mediation to contain privacy-propagation externalities in multi-agent LLM workflows, built on information-flow-control ideas; Evolve judges the concept sound while noting its zero-leak results are partly by construction. Both papers carry medium reproducibility readiness, giving a reproducibility mean of 55 across 2 reviewed papers. The bibliometric field distribution splits evenly (0.5/0.5) between AI security & privacy (LLM agents) and computer security (LLM-agent systems privacy) — effectively one theme under two labels — which yields a focus score of 0 on this index. Collaboration metrics record 12 distinct coauthors and a collaboration score of 51, with Chen Hou, Tao Huang, and Guosen Wu recurring most often; Zheng appears in middle and last author positions without corresponding authorship in the cataloged work.
Collaboration pattern
Guolong Zheng's cataloged work shows a compact collaboration network of 12 distinct coauthors, with Chen Hou the most frequent partner, followed by Tao Huang and Guosen Wu; Jiayang Meng, Xu Yang, Xuechao Yang, and Feng Xia appear once each. Across the two 2026 arXiv preprints he holds middle and last author positions without corresponding authorship, consistent with contributing specialist roles inside a stable team centered on the Chen Hou, Tao Huang, and Guosen Wu cluster.
Research keywords
Highlights
CIPL: exposure-versus-recovery framing of black-box agent privacy leakage
PAPC: event-level platform mediation for privacy propagation
Timeline
- 2026Releases CIPL on arXiv in a middle author position, formalizing exposure-versus-recovery privacy leakage for black-box LLM agents.
- 2026Publishes PAPC on arXiv as last author, proposing event-level platform mediation for privacy propagation in multi-agent LLM workflows.
- 2026Both preprints enter the Evolve review catalog with medium reproducibility readiness, anchoring a two-paper early-stage profile in LLM-agent privacy.
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 are computed from an OpenAlex-based snapshot only; no citation counts, h-index, or full publication list were available.
- The profile rests on 2 arXiv preprints (both 2026) reviewed by Evolve; no peer-reviewed venue records are included in this snapshot.
- Output, impact, and activity axes are unscored because the underlying bibliometric data is insufficient; no substitute numbers are estimated.
- Identity is name-resolved from public bibliographic data via ORCID 0009-0000-4677-1642; 'Guolong Zheng' may correspond to multiple researchers.