HYBRID PERSONALIZATION ARCHITECTURE

摘要

Belief probabilities for ontological concepts within a preference model (108) representing a belief of user preferences are determined. The belief probabilities are based on user feedback (104). A convolution of the belief probabilities and ontology co-occurrence probabilities is performed. A prioritized list (120) from a plurality of documents (116) based on the convolution is generated. In one aspect, the ontology co-occurrence probabilities for ontological concepts within a co-occurrence model (118) are calculated. The ontology co-occurrence probabilities represent a probability of two ontological concepts being associated with a single document of the plurality of documents. In another aspect, the documents describe multimedia content.