PREDICTING USER ACTIVITY IN SOCIAL MEDIA APPLICATIONS

• US 20140258188A1
• Filed: 03/11/2013
• Published: 09/11/2014
• Est. Priority Date: 03/11/2013
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Mahmud, Jalal U., Nichols, Jeffrey W., Chen, Jilin
• Granted Patent: US 9,117,179 B2
• Time in Patent Office: Days
• US Class Current: 706/12
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/04 Inference or reasoning models
• CPC Class Codes: G06Q 10/00 Administration; Management
• CPC Class Codes: G06Q 50/01 Social networking

Abstract

Embodiments of the invention relate to predicting user activity in a social media application. In one embodiment, user activity information is collected for a user from a social media application and activity features of the user are determined, based on the collected activity information for the user. Then a model is created to predict a future activity of the user in the social media application. The model uses the determined activity features of the user and results obtained from running the created model, to determine future activity of the user in the social media application.

Claims

1. A method for predicting user activity in a social media application, the method comprising the steps of:

collecting activity information for a user from a social media application;

determining, by one or more computer processors, activity features of the user, based on the collected activity information for the user;

creating a model to predict a future activity of the user in the social media application, wherein the model uses the determined activity features of the user; and

determining based on results obtained from running the created model, the future activity of the user in the social media application.

2. The method of claim 1, wherein the activity information includes at least one of:

historic activity patterns of the user, activity of the user immediately before the current time, historic activity patterns of one or more users connected to the user, and activity of the one or more users connected to the user immediately before the current time.

3. The method of claim 1, wherein the determined activity features include at least one of:

a frequency of action, a type of action, a temporal distribution of action, and a content of an action.

4. The method of claim 1, wherein the created model is one or more of:

a mathematical model, a machine learning model, or a combination of a mathematical and a machine learning model.

5. The method of claim 1, wherein the future activity of the user is at least one of an activity level of the user or a time before next action by the user, wherein the activity level of the user is a representation of how active the user is and the time before next action by the user is particular to a type of action.

6. The method of claim 1, further comprising:

creating, by one or more computer processors, a plurality of models to predict future activity of the user;

collecting historical data of the user from the social media application;

comparing results from running the plurality of models to the historical data;

determining results from a first model of the plurality of models are more similar to the historical data than results from a second model of the plurality of models; and

determining whether the first model performs more accurately than the second model, based on the compared results.

7. The method of claim 1, wherein the step of creating a model further comprises:

selecting a model type, wherein the model type is one of a mathematical model, a machine learning model, or a combination of a mathematical and a machine learning model; and

determining settings and parameters for the model type, wherein the settings and parameters are one or more of the determined activity features.

8. A computer program product for predicting user activity in a social media application, the computer program product comprising a computer readable storage medium having program code embodied therewith, the program code readable/executable by a computer processor to:

collect activity information for a user from a social media application;

determine activity features of the user, based on the collected activity information for the user;

create a model to predict a future activity of the user in the social media application, wherein the model uses the determined activity features of the user; and

determine, based on results obtained from running the created model, the future activity of the user in the social media application.

9. The computer program product of claim 8, wherein the activity information includes at least one of:

historic activity patterns of the user, activity of the user immediately before the current time, historic activity patterns of one or more users connected to the user, and activity of the one or more users connected to the user immediately before the current time.

10. The computer program product of claim 8, wherein the determined activity features include at least one of:

a frequency of action, a type of action, a temporal distribution of action, and a content of an action.

11. The computer program product of claim 8, wherein the created model is one or more of:

a mathematical model, a machine learning model, or a combination of a mathematical and a machine learning model.

12. The computer program product of claim 8, wherein the future activity of the user is at least one of an activity level of the user or a time before next action by the user, wherein the activity level of the user is a representation of how active the user is and the time before next action by the user is particular to a type of action.

13. The computer program product of claim 8, the program code further being readable/executable by a computer processor to:

create a plurality of models to predict future activity of the user;

collect historical data of the user from the social media application;

compare results from running the plurality of models to the historical data;

determine results from a first model of the plurality of models are more similar to the historical data than results from a second model of the plurality of models; and

determine the first model performs more accurately than the second model, based on the compared results.

14. The computer program product of claim 8, wherein to create a model comprises the program code further being readable/executable by a computer processor to:

select a model type, wherein the model type is one of a mathematical model, a machine learning model, or a combination of a mathematical and a machine learning model; and

determine settings and parameters for the model type, wherein the settings and parameters are one or more of the determined activity features.

15. A computer system for predicting user activity in a social media application, the computer system comprising:

one or more computer processors;

one or more computer-readable tangible storage media;

program instructions stored on the one or more computer-readable tangible storage media for execution by at least one of the one or more computer processors, the program instructions;

collect activity information for a user from a social media application;

determine activity features of the user, based on the collected activity information for the user;

create a model to predict a future activity of the user in the social media application, wherein the model uses the determined activity features of the user; and

determine, based on results obtained from running the created model, the future activity of the user in the social media application.

16. The computer system of claim 15, wherein the activity information includes at least one of:

historic activity patterns of the user, activity of the user immediately before the current time, historic activity patterns of one or more users connected to the user, and activity of the one or more users connected to the user immediately before the current time.

17. The computer system of claim 15, wherein the determined activity features include at least one of:

a frequency of action, a type of action, a temporal distribution of action, and a content of an action.

18. The computer system of claim 15, wherein the future activity of the user is at least one of an activity level of the user or a time before next action by the user, wherein the activity level of the user is a representation of how active the user is and the time before next action by the user is particular to a type of action.

19. The computer system of claim 15, wherein the program instructions:

create a plurality of models to predict future activity of the user;

collect historical data of the user from the social media application;

compare results from running the plurality of models to the historical data;

determine results from a first model of the plurality of models are more similar to the historical data than results from a second model of the plurality of models; and

determine the first model performs more accurately than the second model, based on the compared results.

20. The computer system of claim 15, wherein the program instructions to create a model:

select a model type, wherein the model type is one of a mathematical model, a machine learning model, or a combination of a mathematical and a machine learning model; and

determine settings and parameters for the model type, wherein the settings and parameters are one or more of the determined activity features.