ETH Zurich NLP researcher: tokenisation, prosody–syntax, and topic modelling; 50 works in the last three years
2
papers here
1
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
24
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—
No output score is available yet: the per-work records needed for this axis have not been assembled, so publication volume, venue spread and authorship-position coverage cannot be scored. It would measure how much this researcher produces and in which roles.
Impact—
Impact is not scored yet: although the snapshot lists an h-index of 24 and an i10-index of 40, the citation-level records behind a full impact assessment are not available, so no estimate is offered. This axis would measure citation-based influence.
Focus100/100
Focus scores 100, the maximum. The field distribution assigns this researcher entirely to natural language processing (computational linguistics), and topic counts reinforce the concentration: natural language processing techniques (77), topic modeling (56) and syntax, semantics and linguistic variation (34) dominate, with speech (26) and language neurobiology (18) as coherent extensions.
Collaboration39/100
The collaboration score is 39, built on 6 distinct co-authors captured in this snapshot—a compact recorded network. Recurring partners include Junghyun Min, Alex Warstadt, Tamar I. Regev, Ethan Gotlieb Wilcox, Ahmetcan Yavuz and Clara Meister. In the 2026 catalog entries the researcher is last author on the tokeniser study and middle author on the prosody paper, indicating a mix of leading and supporting roles.
Activity100/100
Activity scores 100, the maximum. The snapshot records 50 works in the most recent three-year window, with annual counts climbing from 1 in 2013 to 11 in 2020, peaking at 20 in 2022 and 24 in 2023, then holding at 19 in both 2024 and 2025, with 12 already recorded in 2026—an intense, continuing publication rhythm.
Reproducibility55/100
The reproducibility score of 55 is the mean over exactly one assessed paper: the 2026 tokeniser study, rated medium reproducibility readiness. Evolve's verdict calls the 2x2 design well-controlled while noting small effects and approximate top-down scoring. With a single reviewed paper, this figure is indicative rather than a robust profile-level estimate.
Research summary
Tiago Pimentel is a natural language processing researcher at ETH Zurich whose profile is name-resolved from OpenAlex bibliographic records tied to ORCID 0000-0002-5159-4641. The snapshot is unusually concentrated: the entire field distribution falls under natural language processing (computational linguistics), with topic mass led by NLP techniques (77), topic modeling (56) and syntax, semantics and linguistic variation (34), extending to speech recognition and synthesis (26) and the neurobiology of language and bilingualism (18). Productivity is the strongest signal: activity scores 100, with 50 works in the most recent three-year window and annual counts rising from 1 (2013) to a 24-work peak (2023). The snapshot also lists an h-index of 24 and an i10-index of 40, though no composite impact score is computed. The Evolve catalog holds two 2026 arXiv papers. In 'Objective vs. Search: Decomposing What Makes a Good Tokeniser' the researcher is last author; Evolve's verdict praises a careful 2x2 study showing the search procedure—not the objective—drives tokeniser bits-per-byte, while noting small effects. 'Using Prosody to Predict Syntactic Structure' lists a middle-author position and awaits review. Six distinct co-authors are recorded, including Clara Meister, Ethan Gotlieb Wilcox and Alex Warstadt.
Collaboration pattern
This snapshot captures a compact network: 6 distinct co-authors and a collaboration score of 39. The recurring circle is consistent—Junghyun Min, Alex Warstadt, Tamar I. Regev, Ethan Gotlieb Wilcox, Ahmetcan Yavuz and Clara Meister. Authorship positions in the 2026 catalog entries range from last author on the tokeniser study to middle author on the prosody paper, suggesting the researcher both leads and contributes within this small, stable group.
Research keywords
Highlights
Decomposing tokeniser quality: search procedure over objective
Prosody as a window onto syntactic structure
The 2026 arXiv paper 'Using Prosody to Predict Syntactic Structure' lists the researcher as a middle author, extending the profile's syntax-and-linguistic-variation thread (34 topic hits) into the prosody–syntax interface. It is not yet reviewed by Evolve, so no verdict, keywords or reproducibility rating are available.
A focused agenda spanning tokenisation to the neurobiology of language
Topic counts sketch a coherent program: natural language processing techniques (77), topic modeling (56), syntax, semantics and linguistic variation (34), speech recognition and synthesis (26), and the neurobiology of language and bilingualism (18)—a wide but connected span that matches the maximum focus score.
Timeline
- 2013Earliest work recorded in the OpenAlex snapshot, with 1 publication that year.
- 2018Output builds to 5 recorded works, up from 2 in 2017, marking the start of steady growth.
- 2020Publication counts accelerate: 11 works in 2020 and 12 in 2021.
- 2023Peak productivity: 20 works in 2022 and 24 in 2023, the highest annual count in the snapshot.
- 2026Both catalog papers appear: the tokeniser decomposition study (last author) and the prosody–syntax paper, with 12 works already recorded this year.
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.
- All bibliometrics come from an OpenAlex snapshot; identity is name-resolved from public bibliographic records tied to ORCID 0000-0002-5159-4641.
- Composite output and impact scores are missing; the h-index (24) and i10-index (40) are quoted verbatim from the snapshot without citation-level records.
- The reproducibility score averages a single reviewed paper, so it should be read as indicative only.
- Only two papers from the Evolve catalog inform this profile; one lacks field, keywords, verdict and reproducibility metadata.