Real-time content integration based on machine learned selections

Patent metadata

Field	Value
Publication number	US-12003577-B2
Application number	US-202318099087-A
Country	US
Kind code	B2
Filing date	Jan 19, 2023
Priority date	May 31, 2017
Publication date	Jun 4, 2024
Grant date	Jun 4, 2024

Abstract

A machine learning engine identifies training data that includes historical user data and historical content data. A machine learning classifier is trained on the training data to generate a relevancy value for each of a plurality of given content items associated with a given user. The relevancy value for each given content item is indicative of a likelihood that the given user will perform a first user device input action and of a likelihood that the given user will perform a second user device input action, in response to being presented with the given content item. The machine learning classifier receives a plurality of candidate content items associated with a first user. The machine learning classifier generates a relevancy value for each candidate content item. At least one of the candidate content items is identified for inclusion in a first content collection based on the generated relevancy values.

First claim

What is claimed is:

1. A method comprising:
identifying training data that includes historical user data and historical content data, at least some of the training data being related to content items that can be selected by performing user device input actions;
training a machine learning classifier on the training data, the machine learning classifier being trained to generate, for each of a plurality of given content items associated with a given user, a relevancy value, the relevancy value for each given content item being indicative of a likelihood that the given user will perform a first user device input action and of a likelihood that the given user will perform a second user device input action, in response to being presented with the given content item as part of a content collection displayed on a user device of the given user, and the first user device input action being different from the second user device input action;
receiving, by the machine learning classifier, a plurality of candidate content items associated with a first user, the candidate content items having been submitted to a platform, each of the candidate content items submitted to the platform with a submitted value that is associated with a bid;
automatically generating, by the machine learning classifier, the relevancy value for each candidate content item associated with the first user; and
identifying at least one of the plurality of candidate content items for inclusion in a first content collection associated with the platform, to be displayed on a first user device of the first user, based on a final value for each of the at least one identified candidate content item, the final value generated based on the submitted value and a normalized relevancy value, the normalized relevancy value being generated by normalizing the relevancy value generated by the machine learning classifier for each of the at least one identified candidate content item.

2. The method of claim 1 , comprising:
generating the final value for each candidate content item, the generating the final value comprising weighting the submitted value for the candidate content item against the normalized relevancy value for the candidate content item.

3. The method of claim 1 , comprising:
generating the final value for each candidate content item, wherein the identifying the at least one of the plurality of candidate content items for inclusion in the first content collection comprises selecting the at least one of the plurality of candidate content items based on the final value generated for each of the at least one candidate content item.

4. The method of claim 1 , comprising:
generating the final value for each candidate content item, wherein the generating the final value for each candidate content item comprises adding the normalized relevancy value to the submitted value for the candidate content item to generate the final value.

5. The method of claim 1 , wherein the machine learning classifier implements a random forest scheme.

6. The method of claim 1 , wherein the machine learning classifier is an ensemble classifier.

7. The method of claim 1 , wherein the training data comprises information specifying whether a given past user of an application associated with the platform selected or bypassed a given content item.

8. The method of claim 1 , wherein the historical user data comprises historical user data of past user actions of past users using an application associated with the platform.

9. The method of claim 8 , wherein the past user actions comprise whether or not a given user installed an application advertised by a given content item.

10. The method of claim 8 , wherein the past user actions comprise browse path data, subscription data, and user profile data.

11. The method of claim 10 , wherein the browse path data describes a browse path of a past user as the past user navigates in the application.

12. The method of claim 10 , wherein the subscription data indicates whether a past user has subscribed to content using the application.

13. The method of claim 10 , wherein the user profile data comprises user preference data of the application.

14. The method of claim 1 , wherein the historical content data comprises metadata that describes content items that have been displayed to past users of an application associated with the platform.

15. The method of claim 1 , wherein the first user device input action and the second user device input action are gestures received through the first user device of the first user, and the first user device is configured to distinguish between the first user device input action and the second user device input action.

16. The method of claim 15 , wherein the first user device input action comprises a tap gesture and the second user device input action comprises a swipe gesture.

17. The method of claim 1 , wherein the first content collection comprises a plurality of content items displayed in sequence and navigable by performing the first user device input action or the second user device input action.

18. The method of claim 1 , wherein the first content collection is generated in response to a request for online content received from the first user device, the request being a request for aggregated content comprising placeholder space and one or more pre-selected content items, and wherein the at least one identified candidate content item is integrated into the placeholder space among the one or more pre-selected content items, the one or more pre-selected content items being selected before the first user device generates the request for the aggregated content.

19. A system comprising:
one or more processors of a machine; and
a memory storing instructions that, when executed by the one or more processors, cause the machine to perform operations comprising:
identifying training data that includes historical user data and historical content data, at least some of the training data being related to content items that can be selected by performing user device input actions;
training a machine learning classifier on the training data, the machine learning classifier being trained to generate, for each of a plurality of given content items associated with a given user, a relevancy value, the relevancy value for each given content item being indicative of a likelihood that the given user will perform a first user device input action and of a likelihood that the given user will perform a second user device input action, in response to being presented with the given content item as part of a content collection displayed on a user device of the given user, and the first user device input action being different from the second user device input action;
receiving, by the machine learning classifier, a plurality of candidate content items associated with a first user, the candidate content items having been submitted to a platform, each of the candidate content items submitted to the platform with a submitted value that is associated with a bid;
automatically generating, by the machine learning classifier, the relevancy value for each candidate content item associated with the first user; and
identifying at least one of the plurality of candidate content items for inclusion in a first content collection associated with the platform, to be displayed on a first user device of the first user based on a final value for each of the at least one identified candidate content item, the final value generated based on the submitted value and a normalize

Assignees

• Snap Inc

Inventors

• Brewer Jason
• Farnham Rodrigo B
• Lue David B
• Stucky-Mack Nicholas J

Classifications

• G06N20/20 Primary

  Ensemble learning · CPC title

• H04L67/10 Primary

  in which an application is distributed across nodes in the network (software deployment G06F8/60; multiprogramming arrangements G06F9/46) · CPC title

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G06N20/00

  Machine learning · CPC title

• H04L67/14

  Session management (for real-time applications in data packet communications networks H04L65/1066) · CPC title