Abstract: Methods and apparatus, including computer program products, for chunked Hypertext Transfer Protocol (HTTP) video cache routing. A method includes, in a client system linked to at least one arbitrary cache node and a group of interconnected computers that include one or more local cache nodes, the client system including at least a processor and a memory, the memory including a browser, loading a player in the browser, and on the player, playing chunked HTTP video requested from the one arbitrary cache node acting as a master cache node, the master cache node requesting a manifest and modifying the manifest so the requests for the actual video are retrieved by the client system from one of the local cache nodes.

Abstract: Methods and apparatus, including computer program products, for scaling video delivery. A method includes, in a corporate network comprising at least a first server linked to a computing system having a video player, receiving an output of a chunked video protocol in the first server, sending the output from the first server via a multicast to a receiver and, in the receiver, processing the output to enable consumption by the video player.

Abstract: Methods and apparatus, including computer program products, for secure video delivery. A method includes providing a cryptographically signed manifest Uniform Resource Locator (URL) to the Hypertext Transfer Protocol (HTTP) manifest format player, the signed URL including at least a validity time span and an identification of a video, in the manifest format player, loading the manifest URL, the manifest URL pointing to an instantiation of a proxy, in the proxy, loading the manifest URL and verifying a validity of the signed URL, in the proxy, retrieving the manifest URL contents, in the proxy, substituting each non-metadata entry in the manifest URL contents with signed URLs either to an underlying video segment or back to an instantiation of the proxy if an entry another manifest, and in the HTTP manifest player, processing the manifest altered with signed URLs as a standard manifest, playing the content.

Abstract: Methods and apparatus, including computer program products, for using speech to text for detecting commercials and aligning edited episodes with transcripts. A method includes, receiving an original video or audio having a transcript, receiving an edited video or audio of the original video or audio, applying a speech-to-text process to the received original video or audio having a transcript, applying a speech-to-text process to the received edited video or audio, and applying an alignment to determine locations of the edits.

Abstract: Methods and apparatus, including computer program products, for aligning video clips to closed caption files. A method includes receiving a video clip, applying speech-to-text to the received video clip, applying an initial alignment of the speech-to-text output to candidate closed caption text in a closed caption file, identifying one or more islands, and outputting the video clip with the aligned closed caption text.

Abstract: Methods and apparatus, including computer program products, for a search for related items using data channels. A method includes, in a computing system having a processor and memory, processing data channels, each data channel defining a set of criteria against which to match items, receiving an input item, and applying the data channels to identify one or more additional items related to the input item.

Abstract: The invention relates to cross-document entity co-reference systems in which naturally occurring entity mentions in a document corpus are analyzed and transformed into name clusters that represent global entities. In a first aspect of the invention, a name variation module analyzes naturally occurring names of entities extracted from the document corpus and provides an initial set of equivalent names that could refer to the same real world entity. In a second aspect of the invention, a disambiguation module takes the initial set of equivalent names and uses an agglomerative clustering algorithm to disambiguate the potentially co-referent named entities.

Abstract: Topic-centric, dynamically-created web pages are compiled from links to multimedia content elements. Keywords are identified within the content elements and used to identify a set of topics related to the content elements. A query string comprising the keywords is executed against the content elements and in response to a request to display information about a topic the query string associated with the topic is executed against the collection of multimedia content elements to identify content elements related to the topic. A dynamically-created web page is displayed that includes computer-executable links to the set of related multimedia content elements.

Abstract: A system improves recognition results. The system receives multimedia data and recognizes the multimedia data based on training data to generate documents. The system receives user augmentation relating to one of the documents or new documents from a user. The system supplements the training data with the user augmentation or new documents and retrains based on the supplemented training data.

Abstract: A method and apparatus for timed tagging of content is featured. The method and apparatus can include the steps of, or structure for, obtaining at least one keyword tag associated with discrete media content; generating a timed segment index of discrete media content, the timed segment index identifying content segments of the discrete media content and corresponding timing boundaries of the content segments; searching the timed segment index for a match to the at least one keyword tag, the match corresponding to at least one of the content segments identified in the segment index; and generating a timed tag index that includes the at least one keyword tag and the timing boundaries corresponding to the least one content segment of the discrete media content containing the match.