Andre-Louis Rochet

—

Unclaimed · AI-structured profileOpenAlex

Emerging researcher in IR/NLP evaluation methodology with a strong reproducibility record

papers here

reviewed by Evolve

first-author

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—

Insufficient data: total publication volume, venue types, and per-year output are not yet available in the indexed record. This axis would measure overall publishing productivity and how it is distributed across years and publication venues.

Impact—

Insufficient data: citation-based influence cannot be scored yet; the snapshot records an h-index of 0 and an i10-index of 0 under sparse coverage. This axis would measure citations, h-index, and the downstream uptake of the work.

Focus0/100

This axis scores 0, reflecting low topical concentration: OpenAlex splits the field distribution evenly, 0.5 for information retrieval / NLP evaluation methodology and 0.5 for general, and five topics appear in the record, led by Topic Modeling (3), followed by Natural Language Processing Techniques (2) and three single-count topics. Current work spans evaluation methodology, time-series forecasting, and applied NLP rather than one narrow lane.

Collaboration22/100

Collaboration scores 22, consistent with a very small network: OpenAlex records 2 distinct co-authors, and the frequent-coauthor list names Elisha Bajemon and Nathan Thierry. Both catalogued 2026 papers place this researcher in the last author position without corresponding-author status, suggesting a contributor role within a compact, stable team rather than broad multi-institution collaboration.

Activity52/100

Activity scores 52, a moderate level grounded in thin but recent data: OpenAlex counts 5 works in the last three years, distributed as 2 publications in 2025 and 3 in 2026, and the two Evolve-reviewed papers are arXiv preprints. The rising 2026 count indicates an active, ongoing pace, though the record is too short to establish a sustained multi-year rhythm.

Reproducibility78/100

Reproducibility scores 78, the strongest axis in this profile: Evolve assessed 2 papers with a mean readiness of 78. The deterministic generative-engine scoring paper is rated high, supported by adversarial falsification gates and an end-to-end validation protocol, while the TW3Cast forecasting router is rated medium: its train-split-only selection is careful but reports no uncertainty on the claimed 3rd-of-130 GIFT-Eval rank.

Research summary

Andre-Louis Rochet is an emerging researcher working across information retrieval / NLP evaluation methodology and applied machine learning, name-resolved from public OpenAlex bibliographic data with no institutional affiliation, country, or ORCID on file. The indexed record is recent and compact: five works in the last three years, split as two publications in 2025 and three in 2026, with OpenAlex topics led by Topic Modeling (3) and Natural Language Processing Techniques (2). Evolve has reviewed two 2026 arXiv papers, both placing the researcher in the last author position. Scoring Without the Engine validates a deterministic, manipulation-resistant content score for generative engines through adversarial falsification gates; Evolve judges the protocol unusually rigorous and honest, notes the surviving score's practical value is modest, and rates reproducibility readiness high. TW3Cast applies a frozen router over lightly fine-tuned foundation models to GIFT-Eval time-series forecasting, claiming 3rd of 130 with train-split-only selection; Evolve finds the methodology careful but flags missing rank uncertainty, rating readiness medium. Reproducibility scores 78, the profile's strongest signal, while collaboration scores 22 on 2 distinct co-authors (Elisha Bajemon, Nathan Thierry), focus scores 0 on an even 0.5/0.5 field split, and output and impact remain unscored pending sufficient data.

Collaboration pattern

The collaboration footprint is narrow: OpenAlex records 2 distinct co-authors, with Elisha Bajemon and Nathan Thierry appearing as frequent co-authors. On both Evolve-reviewed 2026 papers this researcher sits in the last author position without corresponding-author status, a position conventionally associated with senior or supervisory roles within a small, stable team. No multi-institution or consortium-scale collaboration is visible in the indexed record.

Research keywords

generative engine optimizationdeterministic content scoringproxy validationadversarial falsification gatesquery-blindness ceilingweb spam detectionevaluation methodologytime-series forecastingfoundation model routingtopic modelingdomain adaptation and few-shot learningreproducibility

Highlights

Deterministic content score for generative engines, validated end to end
The 2026 paper Scoring Without the Engine builds a manipulation-resistant, deterministic proxy score for generative engines and validates it through adversarial falsification gates. Evolve calls the protocol unusually rigorous and honest, rates reproducibility readiness high, and notes that the surviving score's practical value remains modest.
Frozen router of fine-tuned foundation models for time-series forecasting
TW3Cast (2026) routes across lightly fine-tuned foundation models for GIFT-Eval forecasting, with the router selected entirely on the training split. It claims 3rd of 130; Evolve rates the methodology careful but flags the absence of uncertainty estimates on the rank, assigning medium reproducibility readiness.
A consistent validation-first working pattern
Across the two Evolve-reviewed papers, reproducibility readiness averages 78. Both emphasize explicit validation protocols — falsification gates for engine-proxy scoring and train-split-only selection for forecasting — pointing to a deliberate, verify-first approach to empirical claims, even where the headline results are modest.

Timeline

2025The indexed OpenAlex record begins with 2 publications in this year, preceding the two catalogued 2026 arXiv papers.
2026Indexed output rises to 3 publications this year, bringing the last-three-year total to 5 works.
2026Publishes Scoring Without the Engine, a deterministic, manipulation-resistant content score for generative engines validated end to end; Evolve rates its reproducibility readiness high.
2026Publishes TW3Cast, a frozen router of lightly fine-tuned foundation models for GIFT-Eval forecasting claiming 3rd of 130; Evolve rates its reproducibility readiness medium.

Papers in the Evolve catalog

Frequent co-authors

Elisha Bajemon×1 Nathan Thierry×1

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.

All bibliometrics come from OpenAlex only; the profile is name-resolved from public bibliographic data, with no ORCID, institution, or country provided.
The output and impact dimensions are flagged as missing: the snapshot records an h-index of 0 and an i10-index of 0, so no citation-based score is reported.
Evolve has reviewed only 2 papers while OpenAlex lists 5 works in the last three years, so reviewed coverage may under-represent the full record.
Only the years 2025 and 2026 appear in the indexed record; no career history, awards, or venue-quality data is available.