2026-10-07T04:07:55Z https://www.scidb.cn/oai

10.57760/sciencedb.gfr.00004 2026-09-30T17:01:50Z

2026-09-30 Global patent-to-paper citations doi:10.57760/sciencedb.gfr.00004 en This dataset contains linkage records of patents citing scientific papers through non-patent literature (NPL) references, intended to characterize the flow of knowledge from fundamental research to technological innovation. The data are drawn from the Google Patents Public Datasets (May 2026 release) and OpenAlex (August 2025 release), supplemented with citation records from Reliance on Science (RoS, v64) with a confidence score of 10. The construction procedure is as follows: GROBID is used to parse NPL strings and extract bibliographic metadata such as title, authors, source, and publication date; NPL titles and paper titles are normalized by lowercasing and removing spaces and punctuation, then exactly matched on SHA-256 hash values, with the earliest-published paper retained when several papers share the same title; records are deduplicated at the patent family–paper level, retaining the earliest-published record, and then merged with the RoS data and globally deduplicated. The current public release covers 25 countries and regions, including the United States, China, and the European Union (27 member states counted as a single region), and contains 13,905,635 citation links involving 3,217,712 citing patents (2,527,154 patent families) and 3,861,792 cited papers. Main fields include the publication number, DOCDB family ID, and publication year of the citing patent, as well as the OpenAlex identifier and publication year of the cited paper. The dataset can support research on science–technology linkage, knowledge spillovers, and technology absorption, as well as policy analysis. patent; non-patent literature; scientific paper Mao Jin Tian Xuecan Liang Zhentao PUBLIC https://creativecommons.org/licenses/by/4.0/ dataset Science Data Bank