Optimizing messages to users of a social network using a prediction model that determines likelihood of user performing desired activity

Patent metadata

Field	Value
Publication number	US-9584465-B2
Application number	US-201615163657-A
Country	US
Kind code	B2
Filing date	May 24, 2016
Priority date	May 31, 2012
Publication date	Feb 28, 2017
Grant date	Feb 28, 2017

Abstract

Techniques to optimize messages sent to a user of a social networking system. In one embodiment, information about the user may be collected by the social networking system. The information may be applied to train a model for determining likelihood of a desired action by the user in response to candidate messages that may be provided for the user. The social networking system may provide to the user a message from the candidate messages with a selected likelihood of causing the desired action.

First claim

What is claimed is:

1. A method comprising:
determining a user profile including information describing characteristics of a user of an online system;
logging one or more user activities associated with the user of the online system, the user activities including the user's responses to one or more messages sent to the user, time the user is active on the online system, and a type of activity performed by the user on the online system;
identifying a message response prediction model for the user, the message response prediction model using information from the user profile, the logged one or more user activities associated with the user of the online system and one or more attributes of messages sent to the user, and times when the messages were sent by the online system to the user;
identifying one or more candidate messages, each candidate message including a link associated with a desired activity;
determining a likelihood of the user performing the desired activity included in each candidate message by applying the message response prediction model for the user to each candidate message;
selecting a candidate message based at least in part on the determined likelihoods; and
sending the selected candidate message from the online system to the user at a specified time based at least in part on the determined likelihood.

2. The method of claim 1 , wherein user activities associated with the user of the online system further include activities performed on the online system by other users connected to the user via the online system.

3. The method of claim 1 , wherein characteristics of the user include at least one selected from a group consisting of: user demographics, behavior of the user, behavior of other users connected to the user via the online system, and any combination thereof.

4. The method of claim 3 , wherein the behavior of the user includes at least one selected from a group consisting of: date and time of activities of the user, types of activities of the user, extent of activities of the user, responses of the user to previous messages provided by the online system, and any combination thereof.

5. The method of claim 3 , wherein the behavior of users connected to the user via the online system at least one selected from a group consisting of: date and time of activities of the users connected to the user via the online system, types of activities of the users connected to the user via the online system, extent of activities of the users connected to the user via the online system, and any combination thereof.

6. The method of claim 1 , wherein the characteristics of the user are based on user activities that occurred during a selected period of time.

7. The method of claim 1 , wherein the message response prediction model further uses at least one selected from a group consisting of: a day of the week a candidate message is to be sent, a time the candidate message is to be sent by the online system to the user, and any combination thereof.

8. The method of claim 1 , further comprising updating the message response prediction model for the user with changed information about the user.

9. The method of claim 8 , wherein the updating is performed periodically during a selected interval.

10. The method of claim 8 , wherein the updating is performed continuously during a selected interval.

11. The method of claim 1 , further comprising updating the message response prediction model for the user with action taken by the user in response to the selected candidate message.

12. The method of claim 1 , further comprising generating a message response prediction model for a group of users, including the user, of the online system.

13. The method of claim 1 , further comprising training the message response prediction model using information from the user profile and the user's responses to one or more messages previously sent to the user.

14. A computer program product comprising a non-transitory computer storage medium having instructions encoded thereon that, when executed by a processor, cause the processor to:
determine a user profile including information describing characteristics of a user of an online system;
log one or more user activities associated with the user of the online system, the user activities including the user's responses to one or more messages sent to the user, time the user is active on the online system, and a type of activity performed by the user on the online system;
identify a message response prediction model for the user, the message response prediction model using information from the user profile, the logged one or more user activities associated with the user of the online system and one or more attributes of messages sent to the user, and times when the messages were sent by the online system to the user;
identify one or more candidate messages, each candidate message including a link associated with a desired activity;
determine a likelihood of the user performing the desired activity included in each candidate message by applying the message response prediction model for the user to each candidate message;
select a candidate message based at least in part on the determined likelihoods; and
send the selected candidate message from the online system to the user at a specified time based at least in part on the determined likelihood.

15. The computer program product of claim 14 , wherein user activities associated with the user of the online system further include activities performed on the online system by other users connected to the user via the online system.

16. The computer program product of claim 14 , wherein characteristics of the user include at least one selected from a group consisting of: user demographics, behavior of the user, behavior of other users connected to the user via the online system, and any combination thereof.

17. The computer program product of claim 16 , wherein the behavior of the user includes at least one selected from a group consisting of: date and time of activities of the user, types of activities of the user, extent of activities of the user, responses of the user to previous messages provided by the online system, and any combination thereof.

18. The computer program product of claim 16 , wherein the behavior of users connected to the user via the online system at least one selected from a group consisting of: date and time of activities of the users connected to the user via the online system, types of activities of the users connected to the user via the online system, extent of activities of the users connected to the user via the online system, and any combination thereof.

19. The computer program product of claim 14 , wherein the characteristics of the user are based on user activities that occurred during a selected period of time.

20. A system comprising:
at least one processor; and
a memory storing instructions configured to instruct the at least one processor to:
determine a user profile including information describing characteristics of a user of an online system;
log one or more user activities associated with the user of the online system, the user activities including the user's responses to one or more messages sent to the user, time the user is active on the online system, and a type of activity performed by the user on the online system;
identify a message response prediction model for the user, the message response prediction model using information from the user profile, the logged one or more user activities

Assignees

• Facebook Inc

Inventors

• Arquette Lex
• Chen David Y
• Grewal Emily B
• Moreno Denise
• Ratiu Florin
• Shi Yanxin
• Sidhu Kiranjit Singh
• Weng Ching-Chih
• Yang Huan

Classifications

• G06N7/01

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

• H04L67/306

  User profiles · CPC title

• G06N5/048

  Fuzzy inferencing · CPC title

• H04L51/32 Primary

  Electricity · mapped topic

• H04L67/22

  Electricity · mapped topic