NLP researcher building reproducible retrieval-augmented QA over versioned normative documents at production scale
1
papers here
1
reviewed by Evolve
1
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-09-21) 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—
The underlying publication-count data is not available yet: Evolve has reviewed only one paper by this author, so total output across venues and years cannot be measured. This axis would aggregate paper counts and venue spread over time.
Impact—
No citation or influence metrics are available yet: the input snapshot contains no citation counts for this author or the reviewed paper. This axis would measure citations and downstream uptake of the work.
Focus100/100
Focus scores 100 because the field distribution is perfectly concentrated: fieldDist records a single field — natural language processing (retrieval-augmented generation) — accounting for the one reviewed paper. This indicates a sharply defined research direction around RAG for normative documents, though with only one reviewed paper the concentration reflects a narrow evidence base rather than a long-term pattern; additional papers could sharpen or dilute this picture.
Collaboration28/100
Collaboration scores 28, reflecting a small footprint: the raw data counts three distinct coauthors — Shuaipeng Jin, Jiwei Shi, and Jensen Hsu — tied to the single reviewed paper. The work appears to be a compact-team effort typical of applied, deployment-driven NLP projects, but the sample is one paper, so the score measures breadth of the observed network rather than an established collaboration pattern; broader bibliographic coverage would likely revise it.
Activity—
Publication-frequency data over time is not available, since only one 2026 paper is in scope. This axis would measure publishing cadence and year-over-year research activity.
Reproducibility100/100
Reproducibility scores 100, from reproMean 100 computed over one paper (reproPapers: 1); Evolve independently rates that paper's reproducibility readiness as high. The reviewed system ships with an end-to-end, production-scale evaluation on a 200-question benchmark with LLM-as-a-judge scoring, which supports the strong rating. Evolve's caveat stands, however: the artifact is reproducible as a whole system, so the contribution of the governance mechanism itself is not isolated.
Research summary
Liuyin Wang is an early-stage profile in the Evolve catalog, currently evidenced by a single reviewed paper: a first-author 2026 arXiv preprint on version- and scope-aware question answering over normative documents. The work describes a deployed retrieval-augmented generation (RAG) system evaluated end to end at production scale, reporting 97.7 versus 88.1 against Gemini File Search on its own 200-question normative-document benchmark, with LLM-as-a-judge used for scoring. Evolve's verdict credits reproducibility but notes the artifact is whole-system, so the governance contribution is not isolated. Available bibliometric signals are correspondingly narrow: focus scores 100 because the single reviewed paper falls in one field, natural language processing (retrieval-augmented generation); reproducibility scores 100 (reproMean 100 over one paper, reproducibility readiness high); and collaboration scores 28, with three distinct coauthors — Shuaipeng Jin, Jiwei Shi, and Jensen Hsu. Output, impact, and activity axes lack sufficient data and are left unscored. No institution, country, or ORCID is attached, so this profile is name-resolved from public bibliographic data.
Collaboration pattern
Observed collaboration is narrow: three distinct coauthors (Shuaipeng Jin, Jiwei Shi, Jensen Hsu) on the single reviewed paper, consistent with a small, deployment-oriented team and the low collaboration score of 28. No senior-collaborator or cross-institution pattern can be established from one paper, and no institutional affiliation is on file.
Research keywords
Highlights
Deployed version- and scope-aware QA over normative documents
Governed RAG outranks Gemini File Search on its own benchmark
High reproducibility readiness backed by LLM-as-a-judge evaluation
Timeline
- 2026First-author arXiv preprint on version- and scope-aware question answering over normative documents enters the Evolve catalog.
- 2026The deployed system reports 97.7 versus 88.1 against Gemini File Search on its own 200-question normative-document benchmark.
- 2026Evolve review rates the paper's reproducibility readiness as high, with reproMean 100 over one paper.
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 signals derive from OpenAlex-based data only; output, impact, and activity axes lack sufficient data and are unscored.
- The profile rests on a single reviewed paper (2026 arXiv preprint); every scored dimension is computed from a one-paper evidence base.
- No institution, country, or ORCID is on file; the profile is name-resolved from public bibliographic data and identity confidence is limited.
- Benchmark numbers (97.7 vs 88.1 over 200 questions) come from the authors' own evaluation as summarized in Evolve's review, not from independent verification.