Systems and methods for providing personalized content

Patent metadata

Field	Value
Publication number	US-10320927-B2
Application number	US-201615299035-A
Country	US
Kind code	B2
Filing date	Oct 20, 2016
Priority date	Oct 20, 2016
Publication date	Jun 11, 2019
Grant date	Jun 11, 2019

Abstract

Systems, methods, and non-transitory computer-readable media can generate a set of candidate content items from a plurality of content items that are available in the social networking system, wherein one or more of the candidate content items are to be included in a personalized content stream for a first user. A corresponding score for each of the candidate content items can be generated with respect to the first user. A first set of content items can be determined from the set of candidate content items based at least in part on the respective scores, wherein content items in the first set are included in the personalized content stream.

First claim

What is claimed is:

1. A computer-implemented method comprising:
generating, by a social networking system, a set of candidate content items from a plurality of content items that are available in the social networking system, wherein one or more of the candidate content items are to be included in a personalized content stream for a first user, wherein at least one candidate content item to be included in the personalized content stream is selected based on the at least one candidate content item being posted by a user that is located in a geographic region in which the first user is also located or has visited;
generating, by the social networking system, a corresponding score for each of the candidate content items with respect to the first user; and
determining, by the social networking system, a first set of content items from the set of candidate content items based at least in part on the respective scores, wherein content items in the first set are included in the personalized content stream.

2. The computer-implemented method of claim 1 , wherein generating a respective score for a candidate content item further comprises:
determining, by the social networking system, a likelihood of the first user selecting an option to like a candidate content item through the social networking system, the likelihood being determined using a trained machine learning model, wherein the score for the candidate content item is based at least in part on the likelihood.

3. The computer-implemented method of claim 1 , wherein generating a respective score for a candidate content item further comprises:
determining, by the social networking system, a likelihood of the first user watching one or more additional content items after having viewed a candidate content item, the likelihood being determined using a trained machine learning model, wherein the score for the candidate content item is based at least in part on the likelihood.

4. The computer-implemented method of claim 1 , wherein generating a respective score for a candidate content item further comprises:
determining, by the social networking system, a likelihood of the first user watching a candidate content item to completion, the likelihood being determined using a trained machine learning model, wherein the score for the candidate content item is based at least in part on the likelihood.

5. The computer-implemented method of claim 1 , wherein generating a respective score for a candidate content item further comprises:
determining, by the social networking system, a likelihood of the first user watching a playback of a candidate content item for a duration of time that is longer than an average duration of time the first user watches playback of content items, the likelihood being determined using a trained machine learning model, wherein the score for the candidate content item is based at least in part on the likelihood.

6. The computer-implemented method of claim 1 , wherein generating a respective score for a candidate content item further comprises:
determining, by the social networking system, a likelihood of the first user watching a playback of a candidate content item for a duration of time that is longer than an average duration of time that other users watched playbacks of the candidate content item, the likelihood being determined using a trained machine learning model, wherein the score for the candidate content item is based at least in part on the likelihood.

7. The computer-implemented method of claim 1 , wherein generating the set of candidate content items further comprises:
obtaining, by the social networking system, one or more content items that were liked by at least one second user that the first user is following in the social networking system.

8. The computer-implemented method of claim 1 , wherein generating the set of candidate content items further comprises:
determining, by the social networking system, that the first user has previously liked one or more content items that were posted by at least one second user; and
obtaining, by the social networking system, one or more content items that were liked by the second user.

9. The computer-implemented method of claim 1 , wherein generating the set of candidate content items further comprises:
obtaining, by the social networking system, one or more content items that were posted by users that are located in a geographic region in which the first user is also located or has visited.

10. The computer-implemented method of claim 1 , the method further comprising:
filtering, by the social networking system, the set of candidate content items to exclude content items that are likely to be flagged by users as being inappropriate or content items that were posted by users that have previously been flagged as posters of inappropriate content.

11. A system comprising:
at least one processor; and
a memory storing instructions that, when executed by the at least one processor, cause the system to perform:
generating a set of candidate content items from a plurality of content items that are available in the social networking system, wherein one or more of the candidate content items are to be included in a personalized content stream for a first user, wherein at least one candidate content item to be included in the personalized content stream is selected based on the at least one candidate content item being posted by a user that is located in a geographic region in which the first user is also located or has visited;
generating a corresponding score for each of the candidate content items with respect to the first user; and
determining a first set of content items from the set of candidate content items based at least in part on the respective scores, wherein content items in the first set are included in the personalized content stream.

12. The system of claim 11 , wherein generating a respective score for a candidate content item further causes the system to perform:
determining a likelihood of the first user selecting an option to like a candidate content item through the social networking system, the likelihood being determined using a trained machine learning model, wherein the score for the candidate content item is based at least in part on the likelihood.

13. The system of claim 11 , wherein generating a respective score for a candidate content item further causes the system to perform:
determining a likelihood of the first user watching one or more additional content items after having viewed a candidate content item, the likelihood being determined using a trained machine learning model, wherein the score for the candidate content item is based at least in part on the likelihood.

14. The system of claim 11 , wherein generating a respective score for a candidate content item further causes the system to perform:
determining a likelihood of the first user watching a candidate content item to completion, the likelihood being determined using a trained machine learning model, wherein the score for the candidate content item is based at least in part on the likelihood.

15. The system of claim 11 , wherein generating a respective score for a candidate content item further causes the system to perform:
determining a likelihood of the first user watching a playback of a candidate content item for a duration of time that is longer than an average duration of time the first user watches playback of content items, the likelihood being determined using a trained machine learning model, wherein the score for the candidate content item is based at least in part on the likelihood

Assignees

• Facebook Inc

Inventors

• Zheng Chen
• Dimson Thomas Frederick
• Yang Linji

Classifications

• H04L67/06 Primary

  specially adapted for file transfer, e.g. file transfer protocol [FTP] · CPC title

• H04L51/10

  Multimedia information · CPC title

• G06N20/00

  Machine learning · CPC title

• G06Q30/0241

  Advertisements · CPC title

• G06Q10/10

  Office automation; Time management · CPC title