| US 11,757,972 B2 | ||
| Web traffic routing | ||
| Sean Joey Summers, San Francisco, CA (US); Chritopher James Maddern, New York, NY (US); and Keith Austin Johnson, Brooklyn, NY (US) | ||
| Assigned to Button, Inc., Greenwich, CT (US) | ||
| Filed by Button, Inc., Grenwich, CT (US) | ||
| Filed on Oct. 19, 2022, as Appl. No. 17/969,193. | ||
| Application 17/969,193 is a continuation of application No. 17/358,680, filed on Jun. 25, 2021, granted, now 11,489,908, issued on Nov. 1, 2022. | ||
| Prior Publication US 2023/0164209 A1, May 25, 2023 | ||
| This patent is subject to a terminal disclaimer. | ||
| Int. Cl. H04L 67/02 (2022.01); H04L 45/42 (2022.01); H04L 67/306 (2022.01); H04L 67/63 (2022.01); H04L 29/08 (2006.01); H04L 12/717 (2013.01) | ||
| CPC H04L 67/02 (2013.01) [H04L 45/42 (2013.01); H04L 67/306 (2013.01); H04L 67/63 (2022.05)] | 22 Claims |
|
1. A method for updating a feedback model comprising:
receiving, by one or more computing devices, data corresponding to a set of content requests, the data corresponding to the set of content request including, for each content request in the set of content requests, data characterizing interactions with a respective web resource before and after the respective content request was made; receiving, by the one or more computing devices, responses to each content request of the set of content requests, wherein each response includes one or more actions performed to satisfy the respective content request, wherein the one or more actions were identified by the feedback model in accordance with a respective policy based on at least one objective; and updating, by the one or more computing devices, the feedback model, wherein updating the feedback model comprises:
evaluating one or more metrics to determine actions from the one or more actions of each response that satisfy or fail to satisfy the at least one objective, wherein each of the one or more actions were selected in accordance with the respective policy based on the at least one objective; and
adjusting a mapping between inputs of the feedback model to each of the respective policies based on data corresponding to the determined actions that satisfy or fail to satisfy the at least one objective.
|