Database Overview
To study trends in AI patenting, CSET and 1790 Analytics created a database containing all AI granted patents and published applications (collectively “patent documents”) between January 2000 and March 2020 in all patent systems worldwide. We identified AI patents using a combination of keywords and patent classifications, notably the Cooperative Patent Classification (CPC) and International Patent Classification (IPC). The keywords and patent classifications were adapted from an earlier study of AI patenting carried out by the World Intellectual Property Organization (WIPO), available at https://www.wipo.int/edocs/pubdocs/en/wipo_pub_1055.pdf. In turn, the technology categories in the WIPO study build upon the Computing Classification System (CCS) developed by the Association for Computing Machinery (ACM).
The AI patent database contains a total of 287,532 records. Each record has a single patent document, either a published application, granted, or amended patent. The documents are collected into "patent families." In simple terms, a patent family contains all patent documents associated with the same initial priority application. These may include patent applications across multiple countries, or successive applications in the same country. In total, the AI patent database contains 155,770 patent families.
Additional information is provided on how the patent universe and each of the categories are defined:
- Patent Classifications and Keywords Used to Define AI Patent Universe
- Application Fields Categories with Patent Classifications
- Functional Application Categories with Patent Classifications
- AI Techniques Categories with Patent Classifications and Keywords
The data fields contained within the annotated patent data index are described here.