Government Interest Extraction and Processing

Public data from the United States Patent and Trademark Office (USPTO), the main data source for PatentsView, contain information on “government interest statements” for any patent granted by the USPTO for inventions funded, at least in part, by a federal research grant or government contract. The Bayh-Dole Act of 1980 and its subsequent modifications require patent filers to disclose federal support and rights to the invention on their applications.   

Objective

To enable the analysis of federally supported patenting and subsequent innovation, the PatentsView team developed a text mining algorithm to extract U.S. federal government organizations and their associated contract and award numbers from the government interest statements in the text of the patent. This page describes the algorithm and evaluation process used for extraction.  

Three-Step Process Overview

The following description of the government interest extraction process is organized into three sections:

1. Extraction Process

   Description of the download and preprocessing of government interest records, the method used to extract government organizations and contract numbers, and the cleaning and preparation processes for the final data sets. 

2. Quality Assurance

   Description of the quality assurance (QA) findings from the original QA process in 2016 and the updated QA process conducted in 2020 after government interest process revisions. 

3. Results Analysis

   Basic summary statistics on the government interest data. 

Methodology Documentation

A methods report of the government interest extraction and processing is available to view here: Government Interest Extraction and Processing