Filtering and tailoring multimedia content based on observed user behavior

Abstract

A user is given the ability to control the display of content items such as advertisements, by for example skipping over content items that are not of interest. Metadata tags associated with non-skipped viewings of content are stored and tracked. Subsequently, candidate content items are scored according to their metadata tags, giving higher scores to candidate content items associated with higher occurrences of non-skipped viewings. The higher-scoring candidate content items can then be favored over other candidate content items. Thus, based on the choices the user makes with respect to skipping or not skipping particular content items, inferences are made as to the user's level of interest in various subjects, and subsequent content items are delivered in a personalized manner to the user.

Dependencies are inferred from the claim text.