First-author researcher in efficient, compact and privacy-preserving retrieval for LLM-era search at CUHK-Shenzhen
3
papers here
3
reviewed by Evolve
3
first-author
0
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 insufficient: the snapshot contains no publication-count or productivity series beyond the three cataloged preprints. This axis would measure cumulative and per-year paper volume across venues. No score is estimated.
Impact—
Impact data is insufficient: no citation counts, h-index, or download statistics are available for this profile. This axis would measure citations and uptake of the work by other researchers. No score is estimated.
Focus0/100
The focus score is 0, reflecting a field distribution split exactly into thirds across privacy-preserving information retrieval, information retrieval, and computer networks / web retrieval infrastructure (0.333 each). In practice the three strands are variations on one retrieval agenda — compact hashing, private retrieval, and agent-facing web infrastructure — but bibliometrically the profile reads as evenly spread rather than concentrated in a single primary field.
Collaboration14/100
The collaboration score is 14, based on 1 distinct co-author in the snapshot — Yunming Xiao, who is also the sole frequent collaborator across the catalog. All three 2026 preprints list Hua as first author without corresponding-author flags, consistent with an early, small-team working mode. The pattern suggests close bilateral collaboration rather than a broad network; the scale and diversity of co-authorship cannot be assessed from the current data.
Activity—
Activity data is insufficient: a yearly publication cadence cannot be computed because only 2026 papers are cataloged. This axis would measure ongoing publication rate and recent research engagement. No score is estimated.
Reproducibility55/100
Reproducibility scores 55 (reproMean) across 3 papers with Evolve repro-readiness reviews, each rated medium. The verdicts are consistent: Spruce, the Matryoshka hash paper, and SemDN all show solid measurements or broad ablations, but share caution flags — single-run results and no code release for the hashing work, an untested core coverage mechanism for SemDN, and NDCG-retention plus leakage caveats for Spruce. Releasing code and multi-run evidence is the clearest path upward.
Research summary
Peichun Hua is a researcher at the Chinese University of Hong Kong, Shenzhen, whose work targets the retrieval layer that emerging LLM applications depend on. The Evolve catalog holds three 2026 arXiv preprints, each with Hua as first author: Spruce on privacy-preserving outsourced retrieval, Matryoshka Hash Representations on compact semantic retrieval, and Semantics Delivery Network on web retrieval infrastructure for LLM agents. Together they trace one agenda — making dense retrieval cheaper (compact embeddings, binary hashes), safer (two-server secure multi-party computation, 128-bit hashes, 0.2–3 s query latency for private RAG), and natively suited to agentic access (a CDN-like edge network delivering chunks to LLM agents). Reviewer verdicts credit solid motivating measurements and broad ablations, while flagging untested coverage mechanisms, single-run results, missing code release, and NDCG-retention and leakage caveats — consistent with a reproducibility score of 55 (reproMean) across 3 scored papers. Collaboration is currently narrow: 1 distinct co-author (Yunming Xiao) and a collaboration score of 14. Output, impact, and activity axes lack sufficient bibliometric data and are reported as missing rather than estimated.
Collaboration pattern
Collaboration is narrowly concentrated: the snapshot records exactly 1 distinct co-author, Yunming Xiao, who is also the sole frequent collaborator, and the collaboration score is 14. All three cataloged papers are first-author works without a corresponding-author flag, suggesting a tight two-person or small-team pattern typical of early independent research. Nothing in the data yet indicates lab-scale or cross-institution collaboration.
Research keywords
Highlights
Private RAG retrieval in 0.2–3 s per query with 128-bit hashes
Nested binary hashing beats compact baselines at every tested budget
A CDN-style Semantics Delivery Network for LLM agents
Timeline
- 2026Releases first-author arXiv preprint Spruce on scalable private outsourced retrieval, cutting private retrieval to 0.2–3 s per query via 128-bit hashes under two-server MPC.
- 2026Publishes first-author preprint on Matryoshka hash representations, where two-stage nested binary hashing beats compact baselines at all tested budgets.
- 2026Proposes SemDN, a CDN-style Semantics Delivery Network that rethinks web retrieval infrastructure for LLM agents.
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 dimensions are computed from an OpenAlex-based bibliometric snapshot; the output, impact, and activity axes have insufficient data and are left unscored.
- All cataloged papers are 2026 arXiv preprints; no peer-reviewed venue, citation, or award information is available.
- Reproducibility reflects Evolve repro-readiness reviews of 3 papers (reproMean 55), not community replication.
- Identity is name-resolved from public bibliographic data and the provided ORCID; the co-author picture covers only papers in the Evolve catalog.