IR and NLP researcher at UC Berkeley: agentic deep search, LLM-based evaluation, and computational peer review
3
papers here
3
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
17
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—
Not yet scored: this snapshot does not include a total works count or a full venue list, so overall publication volume cannot be computed. This axis would measure cumulative output and its distribution across venues and years.
Impact—
Not yet scored: although the snapshot reports h-index 17 and i10-index 32, the per-paper citation counts and citation totals needed for an impact score are absent. This axis would measure citation-based influence and uptake of the work.
Focus0/100
Scores 0/100: the field distribution splits evenly at one third each across LLM-based evaluation and scientometrics, scholarly NLP, and agentic deep search, with topic mass spread over information retrieval and search behavior (96), topic modeling (86), and recommender systems (62). Work spans several adjacent threads rather than one narrow lane — breadth over concentration in this snapshot.
Collaboration51/100
Scores 51/100: 12 distinct coauthors appear in the last three years, and the frequent-coauthor list spans 12 named collaborators including Ebrahim Bagheri, Charles L. A. Clarke, Benjamin C. M. Fung, Sajad Ebrahimi, and Shakiba Amirshahi. All three reviewed 2026 papers list this researcher as a middle author rather than corresponding author, consistent with team-based work of moderate network size.
Activity100/100
Scores 100/100: 72 works in the last three years, with yearly counts climbing from 1 (2018) and 2 (2019) through 9 (2021), 14 (2022), 17 (2023), 25 (2024) and 21 (2025) to 26 in 2026, the snapshot's highest. All three arXiv papers in the Evolve catalog are dated 2026, matching the ongoing surge.
Reproducibility70/100
Scores 70/100 across the 3 reviewed papers (reproMean 70). Peerify — the claim-verification benchmark with 800 claims and 6 variants — is rated high readiness, while the LLM-judge reliability analysis and the agentic deep search paper are rated medium, with Evolve noting nested-data caveats and a missing component ablation respectively.
Research summary
Negar Arabzadeh is a researcher at the University of California, Berkeley working across information retrieval and natural language processing, with an OpenAlex-derived profile (identity: openalex) showing an h-index of 17 and an i10-index of 32. Recent output is exceptionally high: 72 works in the last three years, with annual counts rising from 1 in 2018 to 26 in 2026. The topic profile concentrates on information retrieval and search behavior (96), topic modeling (86), and recommender systems (62), with smaller activity in NLP techniques (16) and ethics and social impacts of AI (10). Current field distribution splits evenly across LLM-based evaluation and scientometrics, scholarly NLP, and agentic deep search — each exactly one third. The Evolve-reviewed catalog holds three 2026 arXiv papers, all with middle-author positioning: a reliability analysis finding that 23 of 29 LLM-judge peer-review evaluation metrics conflate style with substance; Question's Gambit, where clue-decomposed pre-loop retrieval lifts Pi-Serini accuracy on BrowseComp-Plus from 83.1 to 90.5 with gpt-5.5; and Peerify, an 800-claim benchmark for verifying peer-review claims against manuscripts. Reproducibility of the reviewed work averages 70/100. Frequent coauthors include Ebrahim Bagheri, Charles L. A. Clarke, and Benjamin C. M. Fung.
Collaboration pattern
Collaboration score 51/100 with 12 distinct coauthors in the last three years. The frequent-coauthor list spans 12 names — including Ebrahim Bagheri, Charles L. A. Clarke, Benjamin C. M. Fung, Sajad Ebrahimi, and Shakiba Amirshahi — suggesting a core recurring group working alongside a rotating set of co-authors. All three Evolve-reviewed papers list middle authorship, not corresponding.
Research keywords
Highlights
LLM review metrics often score style, not substance
Clue decomposition lifts agentic deep search accuracy
Peerify benchmarks peer-review claim verification
Timeline
- 2018Recorded output begins in the OpenAlex-derived counts: 1 work in 2018, followed by 2 in 2019.
- 2021Output accelerates: 9 works in 2021, rising to 14 in 2022.
- 2024Sustained high volume: 17 works in 2023, 25 in 2024, and 21 in 2025; the last three years total 72 works.
- 2026Peak year: 26 works in 2026, including three arXiv papers in the Evolve catalog on LLM-judge reliability, agentic deep search, and peer-review claim verification (Peerify).
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 indicators (h-index, i10-index, yearly counts, topics) come from an OpenAlex-derived snapshot; identity is resolved from public bibliographic data.
- Output and impact dimensions are unscored because total works and citation aggregates were not available in this snapshot.
- The Evolve catalog covers only three 2026 arXiv papers, all with middle (non-corresponding) authorship; other venues and author roles are not represented here.
- The reproducibility score (mean 70) is based on only 3 reviewed papers — two rated medium and one high readiness.