University of Luxembourg researcher using persona-based LLM simulations to augment A/B testing and hypothesis testing.
2
papers here
1
reviewed by Evolve
2
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-17) 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 publication-count metrics needed for this axis are not available in the current snapshot, so no output score is computed. This dimension would measure total and annual publication volume across venues, providing a baseline picture of research productivity.
Impact—
Citation-based impact data is absent from this snapshot, so the impact axis is unscored. It would capture citation counts, h-index-style aggregates, and how often this researcher's papers, including the two 2026 preprints, are referenced by later work.
Focus100/100
The field distribution contains a single recorded field — machine learning & statistics (experimental design) — yielding a perfect focus score of 100. Both papers reinforce this concentration: one develops prediction-powered hypothesis testing, the other persona-conditioned LLM agents for A/B test simulation, so the recorded output forms one tightly defined research program around learning-augmented experimentation.
Collaboration36/100
The snapshot records five distinct co-authors, giving a collaboration score of 36. Repeated collaborators include Saab Mansour, Weronika Łajewska, Leonardo Perelli, Aymen Al Marjani, and Paul Missault. Both cataloged 2026 papers list Benomar as first author and neither as corresponding, and the small collaborator pool suggests a compact team-based working style rather than a broad, loosely coupled network.
Activity—
Recent-activity signals such as per-year publication counts are not yet available, so no activity score is reported. This axis would track publishing cadence over recent years to show whether the current two-paper, single-year burst reflects a sustained rate.
Reproducibility55/100
Reproducibility readiness averages 55 across the single paper assessed so far (reproPapers = 1), giving a score of 55 at the medium tier. Both cataloged preprints carry a medium readiness rating, so the axis currently reflects one paper's assessment and should not yet be read as a stable track record.
Research summary
Ziyad Benomar is a researcher at the University of Luxembourg whose recent work sits at the intersection of machine learning, statistics, and LLM-based human behavior simulation. The Evolve catalog lists two 2026 arXiv preprints on which Benomar is first author. The first, 'Augmented Hypothesis Testing with Persona-Based LLM Simulations,' combines an asymmetric Z-test with GPPI, a transformed variant of prediction-powered inference (PPI++), to exploit LLM-persona predictions inside A/B testing; Evolve judges the theory solid while noting that the empirical gains across four datasets are real but often modest. The second, 'Data-Driven Persona-Conditioned Agents for A/B Test Simulation,' builds data-driven personas that let LLM agents predict A/B test directions at 0.75-0.90 accuracy; Evolve sees a useful pre-screening tool, though inconsistent numbers hurt trust. The bibliometric snapshot shows an extremely concentrated field profile: the recorded field distribution contains only machine learning & statistics (experimental design), the focus score is 100, and five distinct co-authors underpin a collaboration score of 36. Reproducibility readiness averages 55 (medium) on one assessed paper. Because output, impact, and activity axes lack sufficient underlying data, this profile should be read as an early-stage snapshot of a specialized line of work.
Collaboration pattern
Both cataloged 2026 papers are first-author arXiv preprints without a corresponding-author role, and the snapshot lists 5 distinct co-authors (collaboration score 36). The recurring names — Saab Mansour, Weronika Łajewska, Leonardo Perelli, Aymen Al Marjani, Paul Missault — indicate a stable small core team, and the narrow collaborator pool matches the single-field focus, though the modest score suggests limited collaboration breadth beyond this group so far.
Research keywords
Highlights
Theory-grounded LLM-assisted A/B testing
Personas that predict A/B test directions at 0.75-0.90 accuracy
Timeline
- 2026First-author arXiv preprint 'Augmented Hypothesis Testing with Persona-Based LLM Simulations' appears, combining an asymmetric Z-test with GPPI, a transformed PPI++, for LLM-assisted A/B testing.
- 2026Second first-author arXiv preprint 'Data-Driven Persona-Conditioned Agents for A/B Test Simulation' follows, reporting persona-driven LLM agents that predict A/B test directions at 0.75-0.90 accuracy.
- 2026Across the two 2026 preprints, the Evolve-recorded field distribution is a single field — machine learning & statistics (experimental design) — with 5 distinct co-authors and a mean reproducibility readiness 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.
- Dimension scores are computed from a bibliometric snapshot; only three of six axes (collaboration, focus, reproducibility) had enough underlying data to score.
- Output, impact, and activity metrics are missing from the underlying data, so those axes are left unscored rather than estimated.
- No citation counts, h-index, venue list, ORCID, or country data were provided; the profile is name-resolved from public bibliographic data only.
- The Evolve catalog holds just two 2026 arXiv preprints for this researcher, and the reproducibility mean (55) rests on a single assessed paper.