Systems and methods described herein allow for optimization of selection and monetization of ads for presenting during two-way conversations between users and a virtual personal assistant. A data processing system can receive an audio query from a client device and identify a user request based on the received audio query. The data processing system can determine a thread of tasks associated with the identified user request, and determine predicted segment durations for the identified tasks. The data processing system can identify candidate ads for presenting during a given segment, and determine a revenue value for each candidate ads. The data processing system can then select an ad among the candidate ads based on the determined revenues and the predicted segment durations. The data processing system can then transmit the selected ad to the client device for presentation to a corresponding user.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.