System and method for improved content discovery

Abstract

Systems and methods for recommending content in a media system are described. In an embodiment, the method identifies a reference metadata tag associated with content in a database, retrieves a data structure storing information of learned associations between the reference metadata tag and a plurality of other metadata tags, the data structure defining a tree of nodes each associated with a corresponding metadata tag, a root node of the retrieved tree associated with the identified reference metadata tag. A path is extracted from the data structure, spanning from the reference metadata tag to a target metadata tag, the extracted path identifying a sequence of metadata tags defined by nodes along the extracted path. Recommended content items are returned based on the sequence of metadata tags identified by the extracted path.

Dependencies are inferred from the claim text.