NLP researcher in retrieval-augmented generation; first author of cost-aware graph-RAG structure pricing work (EffiRAG)
1
papers here
1
reviewed by Evolve
1
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-09-20) 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—
Insufficient data: no publication-volume or venue metrics are available for this researcher yet. This axis would measure total scholarly output — number of papers, venues, and authorship roles across years — and cannot be scored from the single cataloged paper.
Impact—
Insufficient data: citation counts, h-index and venue influence are not yet available for this profile. This axis would measure how widely the work is cited and adopted; with no citation record in the snapshot, no score can be reported.
Focus100/100
The focus score of 100 reflects complete concentration: the field distribution shows 100% of cataloged work (1 of 1 papers) in natural language processing, specifically retrieval-augmented generation. The 2026 arXiv paper on structure pricing in graph-based RAG, with its cost-quality trade-off analysis, defines the entire known research footprint — a sharply narrowed, single-topic profile at this early stage.
Collaboration36/100
Collaboration scores 36 on the 0-100 scale, with 5 distinct coauthors recorded. Recurring collaborators include Haoyang Ma, Chao Peng, Lionel Briand, Boxi Yu and Jialun Cao. The cataloged 2026 paper is first-authored but not corresponding, suggesting a core-contributor role within a team rather than sole leadership; the small coauthor pool points to a compact, early-stage network.
Activity—
Insufficient data: only one dated publication (2026) exists in the snapshot, so publishing cadence cannot be measured. This axis would track output rhythm — papers per year and continuity of production over time — once more dated records accumulate.
Reproducibility55/100
Reproducibility averages 55 across 1 scored paper, whose Evolve readiness is rated medium. The reviewed work is credited with a strong cost-aware evaluation protocol, judged over 120 queries with LLM-as-a-judge, but its benchmark is small and LLM-generated, leaving artifact and setup completeness as open gaps.
Research summary
Yuzhong Zhang is an early-stage researcher in natural language processing whose cataloged work concentrates on retrieval-augmented generation (RAG), specifically on the economics of graph-based RAG systems. Evolve's catalog contains one first-authored paper, a 2026 arXiv preprint titled When Is Graph Structure Worth Its Cost? The Case for Structure Pricing in Retrieval-Augmented Generation, which proposes structure pricing as a way to decide when graph structure actually pays off. The associated system, EffiRAG, wins 93 of 120 judged queries against a LightRAG-hybrid baseline while costing 57% less, a result Evolve's review attributes to a strong cost-aware evaluation protocol using LLM-as-a-judge. The same review cautions that only a small LLM-generated benchmark supports the claims, and the paper's reproducibility readiness is rated medium, with a reproducibility mean of 55 across one scored paper. Bibliometrically, the profile is narrow because the data is narrow: the field distribution places 100% of cataloged output in NLP/RAG, giving a maximum focus score of 100, while collaboration scores 36 with 5 distinct coauthors — Haoyang Ma, Chao Peng, Lionel Briand, Boxi Yu and Jialun Cao — recurring in the record. Output, impact and activity metrics are not yet available, so this profile should be read as a focused, single-paper snapshot of a researcher positioned in cost-aware graph RAG, not as a mature career record.
Collaboration pattern
The collaboration score of 36 rests on 5 distinct coauthors in the record, with Haoyang Ma, Chao Peng, Lionel Briand, Boxi Yu and Jialun Cao recurring most often. The cataloged 2026 paper lists Zhang as first author but not corresponding author, consistent with a core-contributor role inside a small team rather than an independent group lead. No institutional affiliation is available, so cross-organization ties cannot be assessed.
Research keywords
Highlights
EffiRAG beats LightRAG-hybrid at 57% lower cost
Cost-aware evaluation protocol praised
Small LLM-generated benchmark caps the claims
Timeline
- 2026Posts the first-authored arXiv preprint When Is Graph Structure Worth Its Cost? The Case for Structure Pricing in Retrieval-Augmented Generation.
- 2026Evolve reviews the paper: EffiRAG wins 93/120 judged queries versus LightRAG-hybrid at 57% lower cost, with a medium reproducibility rating.
- 2026Bibliometric snapshot classifies the work under natural language processing (retrieval-augmented generation), recording 5 distinct coauthors and a focus score of 100.
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 a limited public bibliometric snapshot; output, impact and activity metrics are missing, so no citation or volume claims are made.
- The Evolve catalog contains only one reviewed paper for this researcher (2026, arXiv), so all qualitative judgments trace back to a single study.
- No institution, country or ORCID is attached to this record; Yuzhong Zhang is a common name, and the profile is name-resolved from public bibliographic data.
- Per the Evolve verdict, the reviewed paper's benchmark is small and LLM-generated, and its reproducibility readiness is medium (mean 55 on one paper).