TEMPORAL-BASED RECOMMENDATIONS FOR PERSONALIZED USER CONTEXTS AND VIEWING PREFERENCES

摘要

A method includes receiving temporal-based user profile data including a first set of user profile data and a second set of user profile data. The method further includes determining an aggregated profile covariance (APC) and an aggregated profile variance (APV), and then deriving a regression coefficient based on the APC and the APV. In response to receiving an observation of desirable media content for one or more particular timeslots, the method further includes determining an updated APC based on a covariance of the first set of user profile data and the observation of desirable media content, determining an updated APV based on a variance of the first set of user profile data, and generating a prediction of desirable media content based on the updated APC, the updated APV, and the regression coefficient. The prediction of desirable media content corresponds to the one or more particular timeslots.