FEATURE TRANSFORMATION OF EVENT LOGS IN MACHINE LEARNING

• US 20170316343A1
• Filed: 05/02/2016
• Published: 11/02/2017
• Est. Priority Date: 05/02/2016
• Status: Active Grant

• Current Assignee: Yahoo AD Tech LLC
• Original Assignee: AOL Advertising Incorporated (Apollo Global Management, Inc.)
• Inventors: Shamsi, Davood, Holtan, Hans Marius, Tian, Yuan, Wang, Jing
• Granted Patent: US 10,860,949 B2
• Time in Patent Office: Days
• CPC Class Codes: G06F 11/3438 monitoring of user actions tracking the activity of the user H04L67/535
• CPC Class Codes: G06F 11/3476 Data logging G06F11/14, G06F11/2205 take precedence
• CPC Class Codes: G06F 11/3495 for systems
• CPC Class Codes: G06F 2201/865 Monitoring of software
• CPC Class Codes: G06N 20/00 Machine learning

Abstract

Embodiments of the present invention provide systems, methods, and computer storage media directed at transforming event logs into features for use in machine learning. In embodiments, a method may include receiving an event log for a user. The event log can indicate an occurrence of a first event associated with the user. The method can also include generating a feature value for the first event. The feature value can be indicative of an amount of time that has passed since the occurrence of the first event. Based, at least in part, on the feature value, an occurrence of a second event can be predicted utilizing a predictive model. The prediction can then be output to enable targeted content associated with the second event to be delivered to the user. Other embodiments may be described and/or claimed herein.

Claims

1. A computer-implemented method, comprising:

receiving an event log for a user that indicates an occurrence of a first event associated with the user;

generating a feature value for the first event that is indicative of an amount of time that has passed since the occurrence of the first event;

predicting an occurrence of a second event based, at least in part, on the feature value for the first event; and

outputting a result of the prediction to enable targeted content associated with the second event to be delivered to the user.

2. The computer-implemented method of claim 1, wherein the event log for the user also indicates occurrences of a plurality of additional events associated with the user, the method further comprising:

generating a plurality of additional feature values, each of the plurality of additional feature values respectively associated with an additional event of the plurality of additional events, wherein each of the plurality of additional feature values is based on an amount of time that has passed since the occurrence of the respectively associated additional event.

3. The computer-implemented method of claim 1, wherein generating the feature value for the first event comprises applying a decay factor associated with the first event to an event value associated with the first event.

4. The computer-implemented method of claim 2, further comprising:

selecting the decay factor from a plurality of decay factors that are respectively associated with various events.

5. The computer-implemented method of claim 2, wherein the event logs for the user indicate multiple occurrences of the first event associated with the user, and wherein generating the feature value for the first event is also based on a jump factor associated with the first event.

6. The computer-implemented method of claim 4, further comprising:

selecting the jump factor from a plurality of jump factors that are respectively associated with various events.

7. The computer-implemented method of claim 4, wherein generating the feature value is based, at least in part, on a maximum threshold that limits the effect of the multiple occurrences on the feature value.

8. The computer-implemented method of claim 3, wherein generating the feature value is based on the equation:

f_E(T)=Σ

<sub>i;

Ei=E</sub>J_Ee<sup>−

λ

E(T-Ti)</sup>,

where f_E(T) is the feature value for the first event at time T, λ

_E is the decay factor associated with the first event, and J_E is the jump factor associated with the first event.

9. The computer-implemented method of claim 1, wherein predicting the occurrence of the second event based, at least in part, on the feature value for the first event is accomplished utilizing a predictive model that was generated utilizing a training set of feature values that included feature values associated with the first event and a corresponding set of training labels indicative of the occurrence of the second event.

10. The computer-implemented method of claim 9, further comprising:

updating the predictive model utilizing the feature value for the first event and a training label indicative of whether the user completed the second event.

11. One or more computer-readable storage media having instructions stored thereon, which, when executed by one or more processors, cause the one or more processors to:

receive an event log associated with a user of a plurality of users, the event log indicating the occurrence of at least a first event associated with the user;

generate a feature value for the first event based on a decay factor associated with the first event;

generate, utilizing the feature value, a predictive model that correlates the occurrence of the first event with an occurrence of a second event; and

store the predictive model to enable prediction of the second event based at least in part on occurrences of the first event associated with one or more other users.

12. The one or more computer-readable storage media of claim 11, wherein the event log for the user also indicates occurrences of a plurality of additional events associated with the user, and wherein the instructions, when executed by the one or more processors, further cause the one or more processors to:

generate a plurality of additional feature values, each of the plurality of additional feature values respectively associated with an additional event of the plurality of additional events, wherein each of the plurality of additional feature values is based on an amount of time that has passed since the occurrence of the respectively associated additional event, wherein to generate the predictive model further utilizes the plurality of additional feature values to correlate the occurrences of the plurality of additional events with the occurrence of the second event.

13. The one or more computer-readable storage media of claim 11, wherein the event log for the user indicate multiple occurrences of the first event associated with the user, and wherein to generate the feature value for the first event is also based on a jump factor associated with the first event.

14. The one or more computer-readable storage media of claim 13, wherein the instructions, when executed by the one or more processors, further cause the one or more processors to:

select the decay factor from a plurality of decay factors that are respectively associated with various events, and select the jump factor from a plurality of jump factors that are also respectively associated with the various events.

15. The one or more computer-readable storage media of claim 14, wherein to generate the feature value is based, at least in part, on a maximum threshold that limits the effect of the multiple occurrences on the feature value.

16. The one or more computer-readable storage media of claim 13, wherein to generate the feature value is based on the equation:

f_E(T)=Σ

<sub>i;

Ei=E</sub>J_Ee<sup>−

λ

E(T-Ti)</sup>,

where f_E(T) is the feature value for the first event at time T, λ

_E is the decay factor associated with the first event, and J_E is the jump factor associated with the first event.

17. A system, comprising:

one or more processors; and

memory, coupled with the one or more processors, having instructions stored thereon, which, when executed by the one or more processors, cause the one or more processors to;

receive an event log associated with a user of a plurality of users, the event log indicating the occurrence of at least a first event associated with the user;

generate a feature value for the first event based on a decay factor associated with the first event;

generate, utilizing the feature value, a predictive model that correlates the occurrence of the first event with an occurrence of a second event; and

store the predictive model to enable prediction of the second event based at least in part on occurrences of the first event associated with one or more other users.

18. The system of claim 17, wherein the event log is a first event log, the user is a first user, the feature value is a first feature value and wherein the instructions further cause the one or more processors to:

receive a second event log for a second user that indicates an occurrence of a third event for the second user;

generate a second feature value for the third event that is indicative of an amount of time that has passed since the occurrence of the third event for the second user;

predict an occurrence of the second event based, at least in part, on the feature value for the third event; and

output a result of the prediction to enable targeted content associated with the second event to be delivered to the user.

19. The system of claim 17, wherein the event log for the user indicates multiple occurrences of the first event associated with the user, and wherein to generate the feature value for the first event is also based on a jump factor associated with the first event, and wherein the instructions further cause the one or more processors to select the decay factor from a plurality of decay factors that are respectively associated with various events, and select the jump factor from a plurality of jump factors that are also respectively associated with the various events.

20. The system of claim 19, wherein to generate the feature value is based on the equation:

f_E(T)=Σ

<sub>i;

Ei=E</sub>J_Ee<sup>−

λ

E(T-Ti)</sup>,

where f_E(T) is the feature value for the first event at time T, λ

_E is the decay factor associated with the first event, and J_E is the jump factor associated with the first event.