Frequently Asked Questions

PRI (Pepkio Research Index) is an open-research project that identifies and analyzes the most-cited recent papers across scientific disciplines. We use transparent, reproducible methods to surface high-impact research and reveal the trends that define each field.

The "Class of" label refers to the group of papers published in a given calendar year whose 18-month citation window has just closed. For example, the Class of 2026 consists of articles published in 2024 that have been tracked for citations for a full 18 months after publication.

We rank papers by their 18-month citation count—the number of citing works published within 18 months of the paper's publication date. We start with all research articles in the target topic's OpenAlex classification, compute this metric, and then apply a relevance filter to keep only papers that are substantially about the topic. The top-ranking papers are then presented in descending order.

An 18-month window balances two goals: it's long enough for papers to accumulate meaningful early citations, yet short enough to capture the current frontier of a field. It provides a standardized measure of early impact, allowing fair comparison across publication years and disciplines.

All bibliographic data (publication dates, citation counts, abstracts, topic classifications) come from OpenAlex, a free and open scholarly database. Using OpenAlex ensures the rankings are transparent, reproducible, and not reliant on proprietary indexes.

After ranking by 18-month citations, we use a large language model (LLM) to screen each candidate's title and abstract. The classifier asks whether the paper's research is substantively about the target field. Only papers that pass this test enter the final ranking. This step removes articles that are only tangentially related to the core field.

We exclude reviews and other non-research articles so that the ranking compares like-with-like. Original research articles represent the primary advancement of knowledge, and excluding reviews keeps the citation-impact comparison consistent.

We extract scientific concepts from the titles and abstracts of the top-cited papers using an LLM pipeline. These concepts are then normalized into a two-level taxonomy (canonical topics and parent themes). Word clouds and trend comparisons are built from how often each concept appears across papers in the cohort, revealing the dominant themes and year-over-year shifts.

Standard most-cited lists often use lifetime citations and can be dominated by older, well-established work. Our approach uses a fixed early-impact window (18 months) and a rigorous topical-relevance filter, so the list highlights recent, field-specific breakthroughs rather than all-time classics or cross-disciplinary outliers.

Yes. All rankings, figures, and supplementary data on this site are published under a CC BY 4.0 license. You are free to share and adapt the material as long as you give appropriate credit.

We release a new "Class of" cohort once per year, shortly after the 18-month citation window closes for the previous calendar year's publications. Older cohorts remain available for comparison.

The ranking reflects the bibliographic snapshot at the time of data retrieval (listed on the page). We do not retroactively adjust rankings for retractions or post-publication changes. Users should consult the latest version of a paper for its current status.

Our rankings are generated automatically from OpenAlex data and an LLM-based relevance filter. While we take great care, it is possible for a paper to be miscategorised by the source database or incorrectly filtered. If you believe a relevant paper was overlooked, please email hello@pepkio.com or use the feedback button on a ranking page. Include the paper's title and DOI so we can investigate.