Abstract: In some embodiments, a method inputs a sequence of historical behaviors for a plurality of instances of content into a prediction network to generate a sequence of values that model the sequence of historical behaviors. A restriction on an operation performed by the prediction network is based on a characteristic of an viewing behavior. A sequence of attention scores is generated based on a similarity of a current behavior for a first instance of content to respective instances of historical behaviors in the sequence of historical behaviors. The method adjusts respective values based on corresponding attention scores to generate an adjusted sequence of values. The adjusted sequence of features are sampled to generate an output from the prediction network that models the sequence of historical behaviors based on the current behavior. The output for determining a prediction if the current behavior is indicative of the viewing behavior.

Abstract: In some embodiments, a method analyzing a first set of values for a first bin plane in a plurality of bin planes. The plurality of bin planes are used to determine a context model for entropy coding of a current block in a video. The method determines whether to use a second set of values from a second bin plane based on the analyzing. When it is determined to use the second set of values, information is calculated for the context model using the first set of values and the second set of values. When it is determined to not use the second set of values, information is calculated for the context model using the first set of values.

Abstract: In some embodiments, a method receives a message that identifies an asset being viewed on a video service at a timestamp. The message is associated with a profile identifier and a plurality of messages are received for the profile identifier while using the video service. The method determines a channel identifier for a channel in which the asset is offered during the timestamp and determines whether a channel change occurred based on the channel identifier and a prior channel identifier determined from a prior message in the plurality of messages. When the channel change has occurred, the method stores the channel identifier and timestamp for the profile identifier in a data structure, wherein a list of channels is generated by querying the data structure to determine a set of channel identifiers and timestamps for the profile identifier.

Abstract: In some embodiments, during synchronizing of files in a source data set to a destination data set, a method receives a set of events that occurred at the source data set after replicating an image of the source data set to the destination data set. The method analyzes the set of events to determine if an exception to a first set of rules for performing a set of operators on the destination data set for the set of events occurs. A second set of rules for the exception is selected based on analyzing the set of events. The method processes the set of operators for the set of events according to the second set of rules to synchronize data from the first data set to the second data set based on the set of events. The processing of the set of operators uses the second set of rules.

Abstract: In some embodiments, a method receives a request for a video and receives a ranking of content delivery networks for the request. Also, global constraints are received for the content delivery networks. The global constraints define an allocation for respective content delivery networks. The method selects a content delivery network using the global constraints and the ranking. A score for the content delivery network is altered from a first score to a second score in the ranking and used in the selecting of the content delivery network. The content delivery network that is selected for the request is outputted.

Abstract: In some embodiments, a method receives a plurality of video segments of a video using a plurality of computing instances and measures data received for the plurality of computing instances during a time window. A video segment for a computing instance in the plurality of computing instances is not fully downloaded during the time window. The method calculates a video delivery metric for the plurality of computing instances based on the data received and calculates an evaluation measurement based on the video delivery metric. The evaluation measurement is used to control an addition or a termination of a computing instance.

Abstract: In some embodiments, a method receives a history of videos that were viewed on a video delivery system as a first sentence in a sequential order and a target video as a second sentence as input to a prediction network. The prediction network analyzes representations for the history of videos and a representation of the target video. The prediction network generates a session representation based on bidirectionally analyzing a sequence of the first representations and the second representation. The method uses the session representation to determine whether to recommend the target video.

Abstract: In some embodiments, a method receives a time period in which video files will be delivered through a plurality of nodes. The method determines a plurality of user accounts that may use a video delivery system to deliver the video files in the time period; generates probabilities of which nodes may be used for each of the plurality of user accounts in the time period; and generates probabilities of which video files may be accessed by each of the plurality of user accounts. A list of video files is generated for a node based on the probabilities of which video files may be accessed and the probabilities of which nodes may be used for each of the plurality of user accounts. The method causes at least a portion of the list of video files for the node to be stored on the node.

Abstract: In some embodiments, a method receives an supplemental content placement and a context associated with a request for supplemental content to be displayed for the supplemental content placement. A first value is generated based on the context using a prediction network for a platform. The method determines probabilities for a plurality of types of request actions based on the context. Then, a threshold for the supplemental content placement is calculated based on the first value and the probabilities for the plurality of types of request actions. The method submits the threshold to a platform in a request for the platform to submit a second value for the supplemental content placement.

Abstract: In some embodiments, a method receives a request for supplemental content to be provided in association with main content. The method selects an instance of supplemental content based on a long-term reward metric and a short-term reward metric. The long-term reward metric is based on feedback from delivery of a plurality of instances of supplemental content and a delivery status for a delivery constraint of one instance of supplemental content. The short-term reward metric is based on feedback from delivery of the one instance of supplemental content. The long-term reward metric is based on feedback from delivery of a plurality of instances of supplemental content and the short-term reward metric is based on feedback from delivery of one instance of supplemental content. The instance of supplemental content is provided to a client device.

Abstract: In some embodiments, a method selects an account associated with an application that is displaying content. A plurality of contact options is retrieved where the contact options in the plurality of contact options are able to receive pull notifications for the account. The method ranks the plurality of contact options based on information associated with the account and generates a plurality of pull notification options using the ranking of the plurality of contact options. The plurality of pull notification options is output to the application for display on the application. A pull notification option in the plurality of pull notification options is associated with information to cause an associated contact option to be contacted upon selection of the pull notification option.

Abstract: In some embodiments, a method receives a distribution area identifier for media content. A distribution area is defined by one or more geographic primitives and associated with a distribution area identifier. A geographic primitive is defined by one or more location identifiers. The method determines whether the one or more geographic primitives of the media content is applicable to a geographic primitive that is determined from a location identifier associated with a request.

Abstract: In one embodiment, a method receives a first stream identifier for a video stream. The first stream identifier is received in a request from a client for the video stream. The method selects a virtual stream identifier for the first stream identifier where the virtual stream identifier is associated with a plurality of physical stream identifiers associated with a plurality of physical streams. A ranking of the plurality of physical stream identifiers is analyzed to select one of the physical stream identifiers. Then, the method outputs the one of the physical stream identifiers as the video stream to use in response to the request for the first stream identifier. The one of the physical stream identifiers is sent to the client to allow the client to request the respective physical stream.

Abstract: In some embodiments, a method receives a first ranking of content delivery networks for a playback session of a video and receives global constraints for the content delivery networks. The global constraints define an allocation for respective content delivery networks. The method alters a score for a content delivery network in the first ranking to generate a second ranking of content delivery networks. Then, a content delivery network is selected from the second ranking for the playback session of the video.

Abstract: In some embodiments, a method receives a first set of video files at a node that delivers video files to client devices and receives a second set of video files. The second set of video files are predicted to be delivered by the node to a client device during a time period. The node receives a request for a video file from a client device and determines whether the video file is stored in the first set of video files and the second set of video files. When the video file is stored at the node, the node sends the video file from the first set of video files and the second set of video files. When the video file is not stored in the first set of video files and the second set of video files, the node sends a request for the video file to another node.

Abstract: In some embodiments, a method receives a first textual description of content and converts the first textual description of content to a first image representation. The method compares a similarity between the first image representation and a second image representation for candidate metadata. The candidate metadata is associated with a second textual description of content. The method determines whether the first textual description of content is associated with the second textual description of content based on the comparison of similarity of the first image representation and the second image representation.

Abstract: In some embodiments, a method selects a sequence of programs watched by a user account. The method calculates a first set of weights based on comparing content of the sequence of programs to content of a target program and calculates a second set of weights based on an order of the sequence of the programs and the first of weights. The first set of weights and the second set of weights are applied to the sequence of programs to generate a prediction of a similarity of the sequence of programs to the target program. Then, the method outputs the prediction of the similarity for use in determining a recommendation for the user account.

Abstract: In some embodiments, a method sends a first instance of a media presentation description to a client device for a media presentation. The first instance of the media presentation description includes location information for a patch update. The location information for the patch update is received in a request from the client device. The method uses the location information to identify the patch update to the media presentation description and sends a second instance of the media presentation description that includes update information from the patch update to update a stored version of the media presentation description for the client device.

Abstract: In some embodiments, a method reviews information from a set of clients. The information is based on interaction with an application associated with a video delivery system. The method selects ascription events from the information based on a model that defines ascription events that could lead to an outcome for the video delivery system and stores the ascription events into a set of data structures. The ascription events are stored based on respective timing information associated with the respective ascription event and one or more of the ascription events that could lead to the outcome are retrievable based on respective time information for the ascription events.

Abstract: In one embodiment, a method receives a first bundle and a second bundle for a video based on metadata for the video. The method stores the second bundle with a dependency on the first bundle when the second bundle is associated with a first set of control data parameters found in the first bundle that control a first aspect of access to the video. The second bundle is stored with a second set of control data parameters to control a second aspect of access to the video. When the first bundle is selected, the method accesses the first set of control data parameters for the first bundle. When the second bundle is selected, the method accesses the first set of control data parameters in the first bundle via the dependency to the first bundle and the second set of control data parameters for the second bundle.