Method and system for generating references to related video

A video processing system for automatically generating references to related videos, comprising at least one processing unit configured to:

• compare a keyword-context pair for a primary video to a plurality of keyword-context pairings, wherein:
• the keyword-context pair for the primary video comprises:
• a keyword comprising one or more words identified within closed caption text of the primary video, and a context of the keyword, the context comprising program metadata of the primary video,
• the plurality of keyword-context pairings is provided in a knowledge base that is a stored database separate from the primary video, the knowledge base comprising:
• a pre-determined listing of a plurality of known keywords, each keyword comprising no express identification of, and no direct reference to, another video,
• a plurality of known contexts, and
• a plurality of pre-determined rules,
• the plurality of keyword-context pairings stored in the knowledge base pairs each one of the known keywords with one or more of the known contexts,
• each one of the keyword-context pairings stored in the knowledge base is associated with a corresponding one of the pre-determined rules stored in the knowledge base, and
• each one of the pre-determined rules comprises one or more actions that, when performed, identify a reference video from the associated keyword-context pairing, the rules being pre-determined using semantic matching that is based on a contextual meaning of the keyword in the known context of the associated keyword-content pairing, to deduce a reference to another video;
• based on the comparing, determine a match of the keyword-context pair with a matching one of the keyword-context pairings in the listing, wherein the keyword-context pair comprises:
• the keyword identified from the primary video, and the context of the keyword;
• take the one or more actions specified by the rule in the listing associated with the matching one of the keyword-context pairings in the listing;
• obtain, from a result of the one or more actions, information identifying a reference video related to the primary video;
• and
• create an annotation comprising program metadata of the reference video related to the primary video.

At least one non-transitory computer readable storage medium having computer program instructions stored thereon that, when executed by at least one processor, cause the at least one processor to automatically generate references for a primary video to related videos by performing the following operations:

• compare a keyword-context pair for the primary video to a plurality of keyword-context pairings, wherein:
• the keyword-context pair for the primary video comprises:
• a keyword comprising one or more words identified within closed caption text of the primary video, and a context of the keyword, the context comprising program metadata of the primary video,
• the plurality of keyword-context pairings is provided in a knowledge base that is a stored database separate from the primary video, the knowledge base comprising:
• a pre-determined listing of a plurality of known keywords, each keyword comprising no express identification of, and no direct reference to, another video,
• a plurality of known contexts, and
• a plurality of pre-determined rules,
• the plurality of keyword-context pairings stored in the knowledge base pairs each one of the known keywords with one or more of the known contexts,
• each one of the keyword-context pairings stored in the knowledge base is associated with a corresponding one of the pre-determined rules stored in the knowledge base, and
• each one of the pre-determined rules comprises one or more actions that, when performed, identify a reference video from the associated keyword-context pairing, the rules being pre-determined using semantic matching that is based on a contextual meaning of the keyword in the known context of the associated keyword-content pairing, to deduce a reference to another video;
• based on the comparing, determine a match of the keyword-context pair with a matching one of the keyword-context pairings in the listing, wherein the keyword-context pair comprises:
• the keyword identified from the primary video, and the context of the keyword;
• take the one or more actions specified by the rule in the listing associated with the matching one of the keyword-context pairings in the listing;
• obtain, from a result of the one or more actions, information identifying a reference video related to the primary video;
• and
• create an annotation comprising program metadata of the reference video related to the primary video.