The Fung Institute is developing several tools to aid in the discovery of patents around a common technology or application. Leveraging advances in machine learning and automated content analysis, the Patent Discovery Project scans a set of hand-coded “training patents” and then finds a range of similar patents based on the similarity of patent abstracts and patent claims. These tools can be used by researchers to identify a broad population of patents for study from a smaller set of known cases, significantly reducing the search time of researchers in finding related patents.
The first application of the Patent Discovery Project is a tool that investigates the emergence of Clean Technologies in the U.S. Patent data. After identifying patents (as described above), the exploration tool below allows users to filter through and browse patents by cluster, technology, and type of patent assignee.