Patent describes way to use training data to create search results.
The U.S. Patent and Trademark Office has granted patent number 11,539,661 (pdf) to GoDaddy (NYSE: GDDY).
Using a learning algorithm to suggest domain names is closely related to patent number 11,245,665 (pdf) for Training a learning algorithm to suggest domain names, which the USPTO granted earlier this year.
GoDaddy summarizes the invention as:
1) creating training data for a learning algorithm to suggest a domain name to a user using training data gathered from the activities of a plurality of users searching for domain names; 2) training the learning algorithm on the created training data to suggest a domain name; and 3) suggesting a domain name to a user using the learning algorithm trained on the training data.
A registrar’s main goal for domain search is conversion. In many ways, it works like Google search results. It monitors how people respond to search results and uses this data to create future results sets. If no one clicks on the first result in a Google search, it will eventually be demoted. Similarly, if people favor certain types of search results at GoDaddy, those will be promoted, and others will be demoted.
GoDaddy applied for this patent in January 2019, and it was granted today.
It will be interesting to see how this technology is implemented and how it benefits GoDaddy’s customers in the future. Their approach of using machine learning for it seems really interesting.