Advancing new approaches and techniques for more sophisticated AI/ML applications for patent search and to assist innovators is the key mission of Teqmine.
This video, shot in our lab, details hierarchical clustering of some 20,000 drone / uav patents. Teqmine’s AI reads the full-text description of each one of these patents, whether they are 2 pages or more than 1000 pages long, and, by using unsupervised machine learning techniques, divides the patents in clusters based on their key concepts.
It starts from very broad level of clustering, but keeps working towards new, more detailed layers of clustering, until we are satisfied with the level of detail, discovery and exploration.
We’re not currently aware of any other public version of large scale hierarchical clustering of patent data, and feel pretty proud of this!
Most importantly, this demo shows how domain (technology) specialized AI’s will be superior to any other approaches when it comes to support humans to search or to have an AI to help them to invent.
If you’re interested to learn more, get in touch with us!