IPRoduct: A Database of Linked Products-intellectual Property Rights

Project Outcomes Statement

The objective of this project was to investigate the possibility of using “virtual patent marking” (VPM) webpages to build a database that links products to patents. Results suggest that 10 to 15 percent of U.S. patentees report patent marking information online. Statistical analysis suggests that companies are more likely to use VPM if they have a higher likelihood of being infringed, if they pursue an active branding strategy, and if they are in greater need of external financing. Companies that use VPM are active in all industrial sectors, ranging from high-text sectors such as medical devices to medium-to-low-tech sectors such as sporting goods.

The project has also identified technical challenges for the large-scale construction of such a database and has proposed solutions. The major challenges include: the ability to perform a large-scale crawl of the web; the need for a high-precision / high-recall classifier of VPM webpages to sort through the vast amount of noise in the crawled data; and the identification of a typology of data structure to extract automatically patent-product pairs from the crawled webpages. Overall, the idea of using information contained on VPM webpages to build a database of patent-product pairs has proved very appealing and it should be executed at a large scale.

To demonstrate the potential impact of such a database, a small “gold standard” dataset has been produced. It focuses on medical devices, biotechnology and pharmaceutical companies, and has been used to study the ‘real impact’ of federal funding for science. We have extracted the scientific publications cited in the patents protecting the commercial products in the dataset, and identified the funding sources for such publications (with a specific focus on the NIH). Overall, we found that 50 to 60 percent of products by biotechnology and pharmaceutical firms exploit NIH-funded research, and about 12% for medical devices. We also found that support is more prevalent among products by small firms than large firms. Finally, today’s products embed science that was published about 25 years ago, and NIH research may find its way into products faster than non-NIH funded research.