US Patent 11748416 Machine-learning system for servicing queries for digital content

In some aspects, a content-extraction system can receive a query from a client device and generate a result set of digital content responsive to the query. For instance, the content-extraction system can obtain, from a search system, a set of digital content matching one or more keywords. The content-extraction system can exclude digital content items lacking core content, digital content items with duplicative content, or both. In some aspects, the content-extraction system can determine, for one or more remaining digital content items, a content attribute score. The content-extraction system can select, as the result set of digital content, a subset of digital content based on the content attribute scores. The content-extraction system can output the result set to the client device.