Meta Platforms, Inc.
Predicting a level of knowledge that a user of an online system has about a topic associated with a set of content items maintained in the online system

Abstract:

An online system generates a hierarchical taxonomy including multiple levels arranged in order of increasing specificity, in which each level includes one or more topics with which a content item maintained in the online system may be associated. Based on information associated with presentations of content items to online system users, which may indicate the users' familiarity with the content items, the online system predicts a likelihood that a particular user is familiar with information associated with one or more content items associated with a topic. Based at least in part on the predicted likelihood, the online system generates a connection between the user and the topic, in which the connection corresponds to a predicted level of knowledge that the user has about the topic. The online system may later retrieve the user's predicted level of knowledge about the topic (e.g., to select content for presentation to the user).