Mohammad Taher Pilehvar

Cardiff University · United Kingdom

Unclaimed · AI-structured profileORCID OpenAlex

NLP researcher at Cardiff University focused on lexical semantics, topic modeling, and evaluation methodology for LLMs

papers here

reviewed by Evolve

first-author

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—

Insufficient data to score. This axis would measure publication volume and productivity across the career; the snapshot provides annual counts and a recent three-year total of 34 works, but not the complete output record needed for a reliable score.

Impact—

Insufficient data to score. This axis would measure citation-based influence and scholarly uptake; although an h-index of 29 and i10-index of 54 are on record, the fuller citation evidence needed for scoring this dimension is not yet available.

Focus100/100

Scores 100, reflecting extreme concentration: OpenAlex maps the profile to a single field, natural language processing (field weight 1). Topic counts reinforce this — Topic Modeling (146) and Natural Language Processing Techniques (116) dominate, with Sentiment Analysis (48), Explainable AI (26), and Domain Adaptation and Few-Shot Learning (17) as adjacent clusters.

Collaboration41/100

Scores 41, a moderate pattern. The recent snapshot records 7 distinct coauthors against 34 works in three years, and the frequent-coauthor list spans Mohammad Reza Modarres, Armin Tourajmehr, Yadollah Yaghoobzadeh, Yi Zhou, Kiamehr Rezaee, Danushka Bollegala, and Jose Camacho-Collados. The two reviewed 2026 papers place him as a middle author and last author respectively, suggesting selective rather than high-volume co-authorship.

Activity100/100

Scores 100, the top activity rating. Annual output has been continuous since 2013 (2 works), peaked at 30 works in 2021, and moved through 21 (2022), 7 (2023), 10 (2024), and 15 (2025), with 9 already recorded in 2026. The recent three-year window totals 34 works, indicating a sustained and currently rising publishing cadence.

Reproducibility55/100

Scores 55, derived from a reproMean of 55 across only one assessed paper (reproPapers = 1), so the figure is a thin signal. The reviewed WiC study is rated medium for reproducibility readiness, with ablations and human error analysis supporting its claims though significance tests are absent; the CLIN paper is likewise medium.

Research summary

Mohammad Taher Pilehvar is a natural language processing researcher at Cardiff University (GB) whose OpenAlex-resolved profile centers on lexical semantics, topic modeling, and evaluation methodology. His topic footprint is led by Topic Modeling (146 works) and Natural Language Processing Techniques (116), with secondary clusters in Sentiment Analysis and Opinion Mining (48), Explainable AI (26), and Domain Adaptation and Few-Shot Learning (17). Cumulative indicators include an h-index of 29 and an i10-index of 54. Output has been sustained: 34 works over the most recent three-year window, with annual counts ranging from 2 (2013) to a peak of 30 (2021), 15 in 2025, and 9 recorded so far in 2026. Evolve has reviewed two 2026 papers: a study arguing that Word-in-Context difficulty stems from unspecified sense granularity rather than weak disambiguation ability, and CLIN, an objective framework for evaluating creativity in short Persian literary text, where LLM judges track human ratings poorly on subjective dimensions. He appears as a middle author on the former and last author on the latter. The field distribution is highly concentrated — natural language processing is the sole mapped field — and the reproducibility figure currently rests on a single assessed paper, warranting caution.

Collaboration pattern

Collaboration is moderate (dimension score 41): the recent snapshot lists 7 distinct coauthors against 34 works in three years. The recurring partner pool includes Mohammad Reza Modarres, Armin Tourajmehr, Yadollah Yaghoobzadeh, Yi Zhou, Kiamehr Rezaee, Danushka Bollegala, and Jose Camacho-Collados — a mix spanning Persian NLP colleagues and international collaborators. On the two reviewed 2026 papers he sits as a middle author on the WiC study and last author on CLIN, consistent with senior but selective co-authorship.

Research keywords

topic modelinglexical semanticsword sense disambiguationWord-in-Context (WiC)large language modelsLLM-as-judgecreativity evaluationsentiment analysisexplainable artificial intelligencedomain adaptationfew-shot learningPersian NLP / low-resource languages

Highlights

WiC is Not WSD: sense granularity, not disambiguation, drives LLM difficulty
Evolve's review of this 2026 arXiv paper reports that Word-in-Context difficulty stems from unspecified sense granularity; ablations and human error analysis support the claim, though significance tests are absent. Pilehvar appears as a middle author and reproducibility readiness is rated medium.
CLIN: an objective framework for evaluating creativity in short Persian literary text
Reviewed by Evolve in 2026, CLIN shows LLM judges track human ratings poorly on subjective Persian creativity dimensions, while its simple proxies match the top judge; human labels are noisy. Pilehvar is the last author and reproducibility readiness is medium.
Sustained output: 34 works in three years, 2021 peak of 30
OpenAlex counts show continuous production from 2013 (2 works) to a 2021 peak (30), then 21 in 2022, 7 in 2023, 10 in 2024, 15 in 2025, and 9 recorded so far in 2026 — 34 works across the recent three-year window.

Timeline

2013Earliest year in the OpenAlex counts: 2 works recorded.
2021Peak annual output with 30 works recorded, more than any other year in the series.
2023A low point of 7 works, followed by recovery to 10 in 2024.
2025Output rebounds to 15 works; 34 works recorded across the recent three-year window.
2026Two Evolve-reviewed papers appear — the WiC sense-granularity study and the CLIN creativity-evaluation framework — with 9 works recorded so far this year.

Papers in the Evolve catalog

Frequent co-authors

Mohammad Reza Modarres×1 Armin Tourajmehr×1 Yadollah Yaghoobzadeh×1 Yi Zhou×1 Kiamehr Rezaee×1 Danushka Bollegala×1 Jose Camacho-Collados×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.

Bibliometrics come from OpenAlex only and the profile is name-resolved from public bibliographic data; output and impact dimensions lack sufficient underlying data to score.
Quality assessment rests on just two Evolve-reviewed papers, both 2026 arXiv preprints.
The reproducibility score (55) is derived from a single assessed paper (reproPapers = 1) and may shift as more papers are reviewed.
2026 publication counts are partial-year and will grow.