NLP researcher at Monash University spanning scientific document understanding and LLM privacy auditing
2
papers here
2
reviewed by Evolve
0
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 metrics are not available in the current data, so this axis is not scored. It would measure total publication volume and annual productivity; the present evidence base covers only two 2026 arXiv preprints.
Impact—
Citation and influence data are unavailable, so impact is not scored. This axis would measure citation counts, citation velocity, and uptake of the researcher's methods by other work; neither paper in the catalog has verified citation data here.
Focus0/100
The focus score is 0, reflecting an observed field distribution split exactly evenly at 0.5 for natural language processing (scientific document understanding) and 0.5 for privacy auditing / NLP security. The two 2026 catalog papers sit one in each area, so current work spans two distinct research lines rather than concentrating in a single field.
Collaboration39/100
The collaboration score is 39, computed from a sample of 6 distinct coauthors in the observed data. The frequent-coauthor list includes Iryna Gurevych, Gholamreza Haffari, Xingliang Yuan, Ting Yu, Shuo Huang, and Serwar Basch. Both 2026 papers are multi-author efforts with Lizhen Qu in middle and last author positions, neither as corresponding author, suggesting contributing rather than leading roles in this sample.
Activity—
Recent publication cadence cannot be computed from the available data, so activity is not scored. It would measure output recency and sustained research tempo over recent years; only the two 2026 arXiv preprints are observable here.
Reproducibility55/100
The reproducibility score is 55, averaged over 2 papers, both rated medium reproReadiness by Evolve. The conformal privacy auditing paper rests on standard theory with finite-sample guarantees that support reimplementation, while ReGround ships a rebuttal-derived dataset; Evolve notes its author-cited-only gold labels cap what grounding scores can mean, a reproducibility and evaluation caveat.
Research summary
Lizhen Qu is a natural language processing researcher at Monash University whose current work splits evenly across two areas: scientific document understanding (0.5 of the observed field distribution) and privacy auditing / NLP security (0.5). The Evolve catalog contains two 2026 arXiv papers. The first, Conformal Privacy Auditing, applies conformal prediction to LLM re-identification auditing and yields finite-sample ambiguity certificates; Evolve's review calls it sound work on standard theory. The second, ReGround, introduces a rebuttal-derived dataset that grounds reviewer comments in multimodal paper evidence, with solid QA and analysis, though author-cited-only gold labels cap what grounding scores can mean. On Evolve's radar, reproducibility scores 55, averaged over 2 papers both rated medium readiness; collaboration scores 39 with 6 distinct coauthors in the observed sample; and the even two-field split yields a focus score of 0. The output, impact, and activity axes lack sufficient data and are not scored. Frequent coauthors include Iryna Gurevych, Gholamreza Haffari, Xingliang Yuan, Ting Yu, Shuo Huang, and Serwar Basch. This is a full-tier profile built on a small, very recent evidence base, so conclusions about trajectory should be read cautiously.
Collaboration pattern
In the observed sample, Lizhen Qu works in small multi-author teams: both 2026 catalog papers are co-authored, with Qu in the middle position on ReGround and the last position on the conformal privacy auditing paper, neither as corresponding author. The radar records 6 distinct coauthors and a collaboration score of 39. Recurring partners include Iryna Gurevych, Gholamreza Haffari, Xingliang Yuan, Ting Yu, Shuo Huang, and Serwar Basch, spanning the scientific-NLP and privacy-auditing tracks.
Research keywords
Highlights
Conformal prediction brings statistical guarantees to LLM re-identification auditing
ReGround: a rebuttal-derived dataset grounding reviewer comments in evidence
A balanced two-track agenda: scientific NLP plus NLP security
Evolve's field distribution for this profile splits evenly at 0.5 / 0.5 between scientific document understanding and privacy auditing / NLP security, and the two 2026 papers occupy one track each — an early but clear spread across evaluation-oriented NLP and LLM risk auditing.
Timeline
- 2026ReGround released on arXiv: a rebuttal-derived dataset grounding reviewer comments in multimodal paper evidence.
- 2026Conformal Privacy Auditing released on arXiv, applying conformal prediction to LLM re-identification auditing with finite-sample ambiguity certificates.
- 2026Evolve catalog snapshot: two reviewed arXiv preprints split evenly between scientific document understanding and privacy auditing / NLP security, with a reproducibility mean of 55.
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 OpenAlex-derived data only; the output, impact, and activity axes lack sufficient data and are not scored.
- The evidence base is two 2026 arXiv preprints reviewed in the Evolve catalog; the reproducibility score of 55 averages only 2 papers.
- Collaboration metrics reflect a small observed sample (6 distinct coauthors) and may not capture the full co-author network.
- Identity is anchored to ORCID 0000-0002-7764-431X and Monash University; other biographical details are not verified in the input data.