USPTO PatentsView
From AltData.wiki, The Alternative Data Encyclopedia
USPTO PatentsView is a research data platform of the United States Patent and Trademark Office (USPTO) that applies artificial intelligence algorithms, including natural language processing and other machine learning techniques, to enhance raw U.S. patent records and convert them into accessible, research-grade datasets for dissemination, visualization, and analysis.[1]
The initiative is led by the USPTO Office of Policy and International Affairs and its Office of the Chief Economist. Its flagship contribution is large-scale entity resolution: disambiguating inventors, organizations, assignees, and locations across pre-grant publications and granted patents so that patenting activity can be tracked consistently over time.[1] PatentsView is being migrated onto the USPTO Open Data Portal (ODP) at data.uspto.gov as the long-term home for its data downloads and APIs.[2][3]
What It Does
Patent documents contain inconsistencies that limit their use for research: inventor names appear in multiple spellings, organizations change form, and the America Invents Act restructured how applicants are represented in source XML. PatentsView's algorithms resolve these complexities, linking the same entities across different patent documents and correcting structural inconsistencies in the USPTO's source data.[1]
Beyond disambiguation, the platform derives statements of government interest on patents together with their attributes, such as the organizations involved, and maintains data linkages that follow inventor and organizational patenting activity over time, with documented methods and sources.[1]
Data And Methodology
PatentsView uses data derived from USPTO full-text data files for patent grants and applications. The USPTO emphasizes that the processed datasets are provided for research purposes and do not constitute the official USPTO record.[1]
The project publishes detailed documentation of its pipeline, covering disambiguation methods, applicant sorting, government interest detection, and patent classification. A peer-reviewed-oriented description of the platform's data linkage and machine learning approach is also available as an SSRN working paper referenced by the USPTO team.[1]
Products
Data downloads available through the Open Data Portal include Annualized Patent Data files highlighting inventor teams, technology fields, and assignees; Granted Patent Disambiguated Data; Granted Patent Long Text Data; Pre-Grant Publication Disambiguated Data and Long Text Data; and Sorted Patent Data (Beta), which offers improved coverage and accuracy for inventor, applicant, and assignee information. Downloads cover granted patents and pre-grant publications through December 2025 at the time of the most recent updates.[1]
Historically, the legacy patentsview.org website also offered a PatentSearch API, interactive search features, and data visualizations; these functions have been subject to temporary interruptions during the transition to the ODP.[2]
History
PatentsView has operated for years from the standalone domain patentsview.org. In 2026 the team announced a transition notice stating that PatentsView is migrating to the USPTO Open Data Portal to support long-term sustainability, reliability, and broader access to high-quality patent data. A dedicated transition guide on data.uspto.gov instructs users migrating from the legacy website, and the USPTO economic research pages now serve as the central access point for PatentsView information.[2][1]
Users And Use Cases
As a free government resource rather than a commercial vendor, PatentsView serves economics and innovation researchers, policy analysts, and technology intelligence practitioners who need disambiguated U.S. patent data for longitudinal analysis. The annualized files are specifically designed to make patent research accessible to non-programmers, while the full disambiguated datasets support large-scale computational work.[1]