Selecting a list of network user identifiers based on long-term and short-term history data

• US 8,527,526 B1
• Filed: 05/02/2012
• Issued: 09/03/2013
• Est. Priority Date: 05/02/2012
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Liu, Jia, Bai, Yijian, Patil, Manojav, Ravichandran, Deepak, Jiampojamarn, Sittichai, Ponnekanti, Shankar
• Primary Examiner(s): Hu, Jensen
• Application Number:
  US13/462,130
• Time in Patent Office: 489 Days
• Field of Search: 707/751
• US Class Current: 707/751
• CPC Class Codes: G06Q 30/0277 Online advertisement

Abstract

A computerized method, system for, and computer-readable medium operable to select a list of network user identifiers. A processing circuit receives a list of network user identifiers represented by long-term history data indicative of web pages visited prior to a first time and short-term history data indicative of web pages visited after the first time and prior to a second time. Long-term interest categories, corresponding weight values for each long-term interest category, short-term interest categories and corresponding weight values for each short-term interest category are identified. A model comprising the long-term and short-term interest categories is generated based on the weight values of the long-term and short-term interest categories using either arithmetic or harmonic progression. The processing circuit receives a list of candidate network user identifiers and generates a list of similar network user identifiers based on the model.

Claims

1. A computerized method of selecting a list of network user identifiers, the method comprising:

receiving, at a processing circuit, the list of network user identifiers;

receiving, at the processing circuit, long-term history data indicative of web pages visited by one or more of the network user identifiers from the received list of network user identifiers prior to a first point in time and short-term history data indicative of web pages visited by the one or more network user identifiers after the first point in time and prior to a second point in time;

identifying, by the processing circuit;

one or more long-term interest categories associated with the long-term history data,

a weight value for each of the one or more long-term interest categories based at least in part on the long-term history data,

one or more short-term interest categories associated with the short-term history data, and

a weight value for each of the one or more short-term interest categories based at least in part on the short-term history data;

generating, by the processing circuit, a model comprising the one or more long-term and short-term interest categories, wherein the model is generated based in part on the weight values for each of the one or more long-term and short-term interest categories;

receiving, at the processing circuit, a list of candidate network user identifiers different from the list of network user identifiers, history data indicative of web pages visited by one or more of the candidate network user identifiers, and corresponding weight data associated with the history data; and

applying, by the processing circuit, the generated model to the list of candidate network user identifiers to generate a list of similar network user identifiers.

2. The method of claim 1, wherein the first point in time corresponds to a time that is at most twenty-four hours prior to a current time.

3. The method of claim 1, wherein the short-term history data is indicative of web pages visited by the network user identifiers within at most four hours prior to a current time.

4. The method of claim 1, wherein the model is generated by arithmetic progression according to an equation having the form:

C=(Ws*S+Wl*L)/(Ws+Wl),

where a score C is a combination of short-term weight Ws and long-term weight Wl, S is a score of the short-term network activity, and L is a score of the long-term network activity.

5. The method of claim 1, wherein the model is generated by harmonic progression according to an equation having the form:

C=(Ws+Wl)/(Ws/S+Wl/L),

where a score C is a combination of short-term weight Ws and long-term weight Wl, S is a score of the short-term network activity, and L is a score of the long-term network activity.

6. The method of claim 1 further comprising generating display data configured to display the list of similar network user identifiers on a user interface.

7. The method of claim 1, wherein the list of similar network users identifiers are generated based in part on a comparison between the weight values for each of the one or more long-term and short-term interest categories associated with the generated model and the history data and corresponding weight data associated with the candidate network user identifiers.

8. A system for selecting a list of network user identifiers comprising a processing circuit operable to:

receive the list of network user identifiers;

receive long-term history data indicative of web pages visited by one or more of the network user identifiers from the received list of network user identifiers prior to a first point in time and short-term history data indicative of web pages visited by the one or more network user identifiers after the first point in time and prior to a second point in time;

identify;

one or more long-term interest categories associated with the long-term history data,

a weight value for each of the one or more long-term interest categories based at least in part on the long-term history data,

one or more short-term interest categories associated with the short-term history data, and

a weight value for each of the one or more short-term interest categories based at least in part on the short-term history data;

generate a model comprising the one or more long-term and short-term interest categories, wherein the model is generated based in part on the weight values for each of the one or more long-term and short-term interest categories;

receive a list of candidate network user identifiers different from the list of network user identifiers, history data indicative of web pages visited by one or more of the candidate network user identifiers and corresponding weight data associated with the history data; and

apply the generated model to the list of candidate network user identifiers to generate a list of similar network user identifiers.

9. The system of claim 8, wherein the first point in time corresponds to a time that is at most twenty-four hours prior to a current time.

10. The system of claim 8, wherein the short-term history data is indicative of web pages visited by the network user identifiers within at most four hours prior to a current time.

11. The system of claim 8, wherein the model is generated by arithmetic progression according to an equation having the form:

C=(Ws*S+Wl*L)/(Ws+Wl),

where a score C is a combination of short-term weight Ws and long-term weight Wl, S is a score of the short-term network activity, and L is a score of the long-term network activity.

12. The system of claim 8, wherein the model is generated by harmonic progression according to an equation having the form:

C=(Ws+Wl)/(Ws/S+Wl/L),

where a score C is a combination of short-term weight Ws and long-term weight Wl, S is a score of the short-term network activity, and L is a score of the long-term network activity.

13. The system of claim 8, wherein the processing circuit is further operable to generate display data configured to display the list of similar network user identifiers on a user interface.

14. The system of claim 8, wherein the list of similar network users identifiers are generated based in part on a comparison between the weight values for each of the one or more long-term and short-term interest categories associated with the generated model and the history data and corresponding weight data associated with the candidate network user identifiers.

15. A computer-readable storage medium having instructions thereon that cause one or more processors to perform operations, the operations comprising:

receiving the list of network user identifiers;

receiving long-term history data indicative of web pages visited by one or more of the network user identifiers from the received list of network user identifiers prior to a first point in time and short-term history data indicative of web pages visited by the one or more network user identifiers after the first point in time and prior to a second point in time;

identifying;

one or more long-term interest categories associated with the long-term history data,

a weight value for each of the one or more long-term interest categories based at least in part on the long-term history data,

one or more short-term interest categories associated with the short-term history data, and

a weight value for each of the one or more short-term interest categories based at least in part on the short-term history data;

generating a model comprising the one or more long-term and short-term interest categories, wherein the model is generated based in part on the weight values for each of the one or more long-term and short-term interest categories;

receiving a list of candidate network user identifiers different from the list of network user identifiers, history data indicative of web pages visited by one or more of the candidate network user identifiers and corresponding weight data associated with the history data; and

applying the generated model to the list of candidate network user identifiers to generate a list of similar network user identifiers.

16. The computer-readable storage medium of claim 15, wherein the first point in time corresponds to a time that is at most twenty-four hours prior to a current time.

17. The computer-readable storage medium of claim 15, wherein the short-term history data is indicative of web pages visited by the network user identifiers within at most four hours prior to a current time.

18. The computer-readable storage medium of claim 15, wherein the model is generated by arithmetic progression according to an equation having the form:

C=(Ws*S+Wl*L)/(Ws+Wl),

where a score C is a combination of short-term weight Ws and long-term weight Wl, S is a score of the short-term network activity, and L is a score of the long-term network activity.

19. The computer-readable storage medium of claim 15, wherein the model is generated by harmonic progression according to an equation having the form:

C=(Ws+Wl)/(Ws/S+Wl/L),

where a score C is a combination of short-term weight Ws and long-term weight Wl, S is a score of the short-term network activity, and L is a score of the long-term network activity.

20. The computer-readable storage medium of claim 15, wherein the operations further comprise generating display data configured to display the list of similar network user identifiers on a user interface.

21. The computer-readable storage medium of claim 15, wherein the list of similar network users identifiers are generated based in part on a comparison between the weight values for each of the one or more long-term and short-term interest categories associated with the generated model and the history data and corresponding weight data associated with the candidate network user identifiers.