PRESENTING CONTENT TO A SOCIAL NETWORKING SYSTEM USER BASED ON CURRENT RELEVANCE AND FUTURE RELEVANCE OF THE CONTENT TO THE USER

• US 20160328481A1
• Filed: 05/08/2015
• Published: 11/10/2016
• Est. Priority Date: 05/08/2015
• Status: Active Grant

• Current Assignee: Meta Platforms, Inc. (f/k/a Facebook, Inc.)
• Original Assignee: Meta Platforms, Inc. (f/k/a Facebook, Inc.)
• Inventors: Gavish, Yael, Marra, Gregory Matthew
• Granted Patent: US 10,303,727 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/24575 using context
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/248 Presentation of query results
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06Q 50/00 Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism healthcare informatics G16H
• CPC Class Codes: H04L 67/306 User profiles

Abstract

A social networking system generates a feed of content items for presentation to a user based on measures of relevance of the content items to the user at a time when the feed is to be presented and at one or more future times. Content items are associated with value functions representing measures of relevance of the content items to the user over time. Based on the value functions for various content items, the social networking system determines an expected amount of user interaction with the feed when the feed is to be presented as well as predicted amounts of user interaction with the feed at one or more future times. Content items are selected for inclusion in the feed to maximize a combination of the expected amount of user interaction and the predicted amounts of user interaction with the feed.

Claims

1. A computer-implemented method comprising:

identifying candidate content items for presentation to a user of a social networking system via a feed of content items at a current time;

obtaining a value function for each candidate content item, the value function for a candidate content item representing a measure of relevance of the candidate content item at different times;

for each of a plurality of combinations of candidate content items included in the feed of content items at the current time;

determining an expected amount of user interaction with the feed of candidate content items at the current time by aggregating measures of relevance for the candidate content items included in the combination of candidate content items based at least in part on the value functions for each of the candidate content items included in the combination, and

determining predicted amounts of user interaction with one or more additional feeds of content items for presentation to the user at one or more additional times subsequent to the current time based at least in part on the combination of candidate content items included in the feed of content items at the current time, a predicted amount of user interaction with an additional feed for presentation to the user at an additional time based at least in part on an aggregation of measures of relevance for candidate content items included in the additional feed at the additional time based at least in part on the value functions for each of the candidate content items included in the additional feed for presentation to the user at the additional time; and

selecting a combination of the plurality of combinations of candidate content items for inclusion in the feed of content items at the current time that maximizes an aggregate of the expected amount of user interaction with the feed of content items at the current time and the predicted amounts of user interaction with the one or more additional feeds of content items for presentation to the user at the one or more additional times subsequent to the current time.

2. The computer-implemented method of claim 1, wherein the value function of a candidate content item varies over time at a rate and the value function of an additional candidate content item varies over time at an additional rate that differs from the rate.

3. The computer-implemented method of claim 1, wherein obtaining the value function comprises retrieving the value function associated with a candidate content item, the value function specified by a user providing the candidate content item to the social networking system.

4. The computer-implemented method of claim 1, wherein obtaining the value function for each candidate content item comprises:

identifying characteristics of a candidate content item;

identifying prior interactions by the user with additional content items having at least a threshold number of characteristics matching the characteristics of the candidate content item; and

determining the value function of the candidate content item based at least in part on the identified prior interactions.

5. The computer-implemented method of claim 1, wherein the value function of a candidate content item is less than a threshold value at a time subsequent to the current time.

6. The computer-implemented method of claim 5, wherein the time subsequent to the current time at which the value function of the candidate content item is less than the threshold value is specified by a user providing the candidate content item to the social networking system.

7. The computer-implemented method of claim 5, wherein the time subsequent to the current time at which the value function of the candidate content item is less than the threshold value is determined by the social networking system based at least in part on characteristics of the candidate content item and prior interactions by the user with additional content items having at least a threshold number of characteristics matching the characteristics of the candidate content item.

8. The computer-implemented method of claim 1, wherein selecting the combination of the plurality of combinations of candidate content items for inclusion in the feed of content items at the current time that maximizes the aggregate of the expected amount of user interaction with the feed of content items at the current time and the predicted amounts of user interaction with the one or more additional feeds of content items for presentation to the user at the one or more additional times subsequent to the current time comprises:

associating weights with each predicted amount of user interaction with an additional feed of content items, a weight associated with the additional feed of content items based at least in part on a difference between the current time and an additional time associated with the additional feeds;

applying the weights to the predicted amounts of user interaction with the one or more additional feeds of content items;

for each of the plurality of combinations of candidate content items included in the feed of content items at the current time, generating an aggregation of an expected amount of user interaction with the feed of content items including the combination of candidate content items at the current time and the predicted amounts of user interaction with the one or more additional feeds of content items for presentation to the user at the one or more additional times subsequent to the current time that include the combination of candidate content items after application of the weights.

9. The computer-implemented method of claim 1, wherein selecting the combination of the plurality of combinations of candidate content items for inclusion in the feed of content items at the current time that maximizes the aggregate of the expected amount of user interaction with the feed of content items at the current time and the predicted amounts of user interaction with the one or more additional feeds of content items for presentation to the user at the one or more additional times subsequent to the current time comprises:

determining a frequency with which the user accesses the feed of content items provided by the social networking system is less than a threshold frequency;

responsive to the determining, identifying combinations of candidate content items including at least one candidate content item having a value function with a measure of relevance less than a threshold within a time interval from the current time, the time interval associated with the threshold frequency; and

selecting a combination from the identified combinations of candidate content items for inclusion in the feed of content items at the current time that maximizes the aggregate of the expected amount of user interaction with the feed of content items at the current time and the predicted amounts of user interaction with the one or more additional feeds of content items for presentation to the user at the one or more additional times subsequent to the current time.

10. The computer-implemented method of claim 1, wherein selecting the combination of the plurality of combinations of candidate content items for inclusion in the feed of content items at the current time that maximizes the aggregate of the expected amount of user interaction with the feed of content items at the current time and the predicted amounts of user interaction with the one or more additional feeds of content items for presentation to the user at the one or more additional times subsequent to the current time comprises:

determining a likelihood of the user accessing the feed of content items between the current time and a time when a value function of a candidate content item has less than a threshold measure of relevance based at least in part on prior accesses of the feed of content items and times associated with the accesses of the feed of content items associated with the user by the social networking system;

responsive to the determined likelihood being less than a threshold likelihood, identifying combinations of candidate content items including the candidate content item; and

selecting a combination from the identified combinations of candidate content items for inclusion in the feed of content items at the current time that maximizes the aggregate of the expected amount of user interaction with the feed of content items at the current time and the predicted amounts of user interaction with the one or more additional feeds of content items for presentation to the user at the one or more additional times subsequent to the current time.

11. A computer program product comprising a computer readable storage medium having instructions encoded thereon that, when executed by a processor, cause the processor to:

identify candidate content items for presentation to a user of a social networking system via a feed of content items at a current time;

obtain a value function for each candidate content item, the value function for a candidate content item representing a measure of relevance of the candidate content item at different times;

for each of a plurality of combinations of candidate content items included in the feed of content items at the current time;

determine an expected amount of user interaction with the feed of candidate content items at the current time by aggregating measures of relevance for the candidate content items included in the combination of candidate content items based at least in part on the value functions for each of the candidate content items included in the combination, and

determine predicted amounts of user interaction with one or more additional feeds of content items for presentation to the user at one or more additional times subsequent to the current time based at least in part on the combination of candidate content items included in the feed of content items at the current time, a predicted amount of user interaction with an additional feed for presentation to the user at an additional time based at least in part on an aggregation of measures of relevance for candidate content items included in the additional feed at the additional time based at least in part on the value functions for each of the candidate content items included in the additional feed for presentation to the user at the additional time; and

select a combination of the plurality of combinations of candidate content items for inclusion in the feed of content items at the current time that maximizes an aggregate of the expected amount of user interaction with the feed of content items at the current time and the predicted amounts of user interaction with the one or more additional feeds of content items for presentation to the user at the one or more additional times subsequent to the current time.

12. The computer program product of claim 11, wherein the value function of a candidate content item varies over time at a rate and the value function of an additional candidate content item varies over time at an additional rate that differs from the rate.

13. The computer program product of claim 11, wherein obtaining the value function comprises retrieving the value function associated with a candidate content item, the value function specified by a user providing the candidate content item to the social networking system.

14. The computer program product of claim 11, wherein obtaining the value function for each candidate content item comprises:

identifying characteristics of a candidate content item;

identifying prior interactions by the user with additional content items having at least a threshold number of characteristics matching the characteristics of the candidate content item; and

determining the value function of the candidate content item based at least in part on the identified prior interactions.

15. The computer program product of claim 11, wherein the value function of a candidate content item is less than a threshold value at a time subsequent to the current time

16. The computer program product of claim 15, wherein the time subsequent to the current time at which the value function of the candidate content item is less than the threshold value is specified by a user providing the candidate content item to the social networking system.

17. The computer program product of claim 15, wherein the time subsequent to the current time at which the value function of the candidate content item is less than the threshold value is determined by the social networking system based at least in part on characteristics of the candidate content item and prior interactions by the user with additional content items having at least a threshold number of characteristics matching the characteristics of the candidate content item.

18. The computer program product of claim 11, wherein select the combination of the plurality of combinations of candidate content items for inclusion in the feed of content items at the current time that maximizes the aggregate of the expected amount of user interaction with the feed of content items at the current time and the predicted amounts of user interaction with the one or more additional feeds of content items for presentation to the user at the one or more additional times subsequent to the current time comprises:

associate weights with each predicted amount of user interaction with an additional feed of content items, a weight associated with the additional feed of content items based at least in part on a difference between the current time and an additional time associated with the additional feeds;

apply the weights to the predicted amounts of user interaction with the one or more additional feeds of content items;

for each of the plurality of combinations of candidate content items included in the feed of content items at the current time, generate an aggregation of an expected amount of user interaction with the feed of content items including the combination of candidate content items at the current time and the predicted amounts of user interaction with the one or more additional feeds of content items for presentation to the user at the one or more additional times subsequent to the current time that include the combination of candidate content items after application of the weights.

19. The computer program product of claim 11, wherein select the combination of the plurality of combinations of candidate content items for inclusion in the feed of content items at the current time that maximizes the aggregate of the expected amount of user interaction with the feed of content items at the current time and the predicted amounts of user interaction with the one or more additional feeds of content items for presentation to the user at the one or more additional times subsequent to the current time comprises:

determine a frequency with which the user accesses the feed of content items provided by the social networking system is less than a threshold frequency;

responsive to the determining, identify combinations of candidate content items including at least one candidate content item having a value function with a measure of relevance less than a threshold within a time interval from the current time, the time interval associated with the threshold frequency; and

select a combination from the identified combinations of candidate content items for inclusion in the feed of content items at the current time that maximizes the aggregate of the expected amount of user interaction with the feed of content items at the current time and the predicted amounts of user interaction with the one or more additional feeds of content items for presentation to the user at the one or more additional times subsequent to the current time.

20. The computer program product of claim 11, wherein select the combination of the plurality of combinations of candidate content items for inclusion in the feed of content items at the current time that maximizes the aggregate of the expected amount of user interaction with the feed of content items at the current time and the predicted amounts of user interaction with the one or more additional feeds of content items for presentation to the user at the one or more additional times subsequent to the current time comprises:

determine a likelihood of the user accessing the feed of content items between the current time and a time when a value function of a candidate content item has less than a threshold measure of relevance based at least in part on prior accesses of the feed of content items and times associated with the accesses of the feed of content items associated with the user by the social networking system;

responsive to the determined likelihood being less than a threshold likelihood, identify combinations of candidate content items including the candidate content item; and

select a combination from the identified combinations of candidate content items for inclusion in the feed of content items at the current time that maximizes the aggregate of the expected amount of user interaction with the feed of content items at the current time and the predicted amounts of user interaction with the one or more additional feeds of content items for presentation to the user at the one or more additional times subsequent to the current time.