Abstract: Described herein is a technique to facilitate filtering during candidate retrieval stage performed by an information retrieval system that utilizes embedding models. An aNN indexing structure is created for each end-user, and in some instances, each activity type. This allows a single request for candidate content items to invoke a single process to obtain content items that satisfy the filtering criteria (e.g., in this case, the in-network requirement) from the separate per-end-user indexes.

Abstract: Described herein is a technique to facilitate filtering during candidate retrieval stage performed by an information retrieval system that utilizes embedding models. An aNN indexing structure is created for each end-user, and in some instances, each activity type. This allows a single request for candidate content items to invoke a single process to obtain content items that satisfy the filtering criteria (e.g., in this case, the in-network requirement) from the separate per-end-user indexes.

Abstract: Apparatus, computer-readable medium, and computer-implemented method for dynamic binding of tasks in a data exchange, including generating an execution plan for the data exchange comprising a plurality of streams, each stream corresponding to a pairing of a producer task and a consumer task and comprising a producer task identifier and a consumer task identifier, binding each producer task to a node at execution time by associating a corresponding producer task identifier with a corresponding node identifier of the node, binding each consumer task to a node at execution time by associating a corresponding consumer task identifier with a node identifier of the corresponding node, and routing an output of each producer task from a node bound to the producer task to a node bound to a paired consumer task by writing the output to a stream corresponding to the pairing of the producer task and the consumer task.

Abstract: Apparatus, computer-readable medium, and computer-implemented method for dynamic binding of tasks in a data exchange, including generating an execution plan for the data exchange comprising a plurality of streams, each stream corresponding to a pairing of a producer task and a consumer task and comprising a producer task identifier and a consumer task identifier, binding each producer task to a node at execution time by associating a corresponding producer task identifier with a corresponding node identifier of the node, binding each consumer task to a node at execution time by associating a corresponding consumer task identifier with a node identifier of the corresponding node, and routing an output of each producer task from a node bound to the producer task to a node bound to a paired consumer task by writing the output to a stream corresponding to the pairing of the producer task and the consumer task.

Abstract: An apparatus, computer-readable medium, and computer-implemented method for performing a data exchange, including receiving tasks for execution, generating an execution plan for executing the tasks on a plurality of nodes, the execution plan comprising one or more data exchanges, each data exchange comprising at least one stream, and each stream identifying a producer task and a consumer task and being configured to transmit output of the producer task as input to the consumer task, executing one or more producer tasks on one or more first nodes in the plurality of nodes based at least in part on the execution plan, and transmitting an output of the one or more producer tasks from the one or more first nodes to one or more streams of the data exchange via a stream application programming interface (API).

Abstract: An apparatus, computer-readable medium, and computer-implemented method for performing a data exchange, including receiving tasks for execution, generating an execution plan for executing the tasks on a plurality of nodes, the execution plan comprising one or more data exchanges, each data exchange comprising at least one stream, and each stream identifying a producer task and a consumer task and being configured to transmit output of the producer task as input to the consumer task, executing one or more producer tasks on one or more first nodes in the plurality of nodes based at least in part on the execution plan, and transmitting an output of the one or more producer tasks from the one or more first nodes to one or more streams of the data exchange via a stream application programming interface (API).