Abstract: Systems and methods are directed to selecting and delivering targeted content. In response to a request from a user, a central control component accesses content delivery data and content provider settings. Based on the content delivery data and the content provider settings, the central control component generates content delivery settings that include a plurality of frequency capping (Fcap) rules. The content delivery settings are transmitted to a serving system. A determination component of the serving system accesses user data associated with the user that indicates user preferences. Based on the content delivery settings and the user data, the determination component selects a piece of content to deliver to the user. A delivery component then delivers the piece of content to the user.

Abstract: In an example embodiment, a solution is provided that enables end-to-end automation across many different components in an online network. End-to-end optimization is a machine learning approach where the entire system, from input to output, is optimized as a whole, without breaking it down into separate components. In other words, the optimization is performed over the entire pipeline of the system, rather than optimizing each component separately.

Abstract: Embodiments of the disclosed technologies receive a first signal from a user session and create a first ranked list of content items. Based on the first signal, a first subset of the first ranked list is assigned to a first set of slots of the user session. A second subset of the first ranked list is assigned to a second set of slots of the user session. Based on user activity data and position context data, a second ranked list of content items is created. The second ranked list is assigned instead of the second subset to the second plurality of slots.

Abstract: Embodiments of the disclosed technologies receive a first signal from a user session and create a first ranked list of content items. Based on the first signal, a first subset of the first ranked list is assigned to a first set of slots of the user session. A second subset of the first ranked list is assigned to a second set of slots of the user session. Based on user activity data and position context data, a second ranked list of content items is created. The second ranked list is assigned instead of the second subset to the second plurality of slots.