System and process for improving recommendations for use in providing personalized advertisements to retail customers

• US 20090281895A1
• Filed: 05/06/2008
• Published: 11/12/2009
• Est. Priority Date: 05/06/2008
• Status: Active Grant

• Current Assignee: RichRelevance Incorporated
• Original Assignee: RichRelevance Incorporated
• Inventors: Selinger, David, Ahuja, Sundeep, Sunwoo, Albert, Osial, James, Kohn, Tyler, DeCourcey, Michael
• Granted Patent: US 8,583,524 B2
• Time in Patent Office: Days
• US Class Current: 705/14.430
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0244 Optimization
• CPC Class Codes: G06Q 30/0246 Traffic
• CPC Class Codes: G06Q 30/0251 Targeted advertisements

Abstract

A method of improving user set recommendations for product advertising including receiving a request for user set recommendations from any of a set of retailers where such request is related to one or more products, receiving from a plurality of user sets from one or more automated user recommendation systems, wherein the plurality of user sets are generated using different user models and using ensemble learning to select one or more most relevant user sets from the plurality of user sets.

Claims

1. A method of improving user set recommendations for product advertising comprising the steps of:

receiving a request for user set recommendations from any of a set of retailers where such request is related to one or more products;

receiving from a plurality of user sets from one or more automated user recommendation systems, wherein the plurality of user sets are generated using different user models; and

using ensemble learning to select one or more most relevant user sets from the plurality of user sets.

2. The process of claim 1 wherein the ensemble learning is based, at least in part, on user behavior collected from user activity subsequent to previous recommendations.

3. The process of claim 2 wherein the ensemble learning utilizes a naï

ve Bayesian network.

4. The process of claim 3 wherein the naï

ve Bayesian network utilizes information concerning the particular retailer requesting recommendations, the page on which the recommendation is to be displayed and the one or more products.

5. The process of claim 4 wherein an observed strength and reliability value for the selection of recommendation sets is used and combined with a strength and reliability value predicted from the naï

ve Bayesian network.

6. The process of claim 1 further including the step of applying one or more filters to one or more of the recommendation sets.

7. The process of claim 6 wherein the step of applying filters uses ensemble learning to select the one or more filters to use.

8. The process of claim 1 including the additional step of communicating the selected user sets electronically.

9. The process of claim 8 wherein the users in the selected user sets are provided with product advertising on a retail website.

10. The process of claim 8 wherein the users in the selected user sets are provided with product advertising not on a retail website.

11. A computer-readable medium having computer-executable instructions for improving user set recommendations for product advertising, the computer-executable instructions causing the system to perform the steps of:

receiving a request for user set recommendations from any of a set of retailers where such request is related to one or more products;

receiving from a plurality of user sets from one or more automated user recommendation systems, wherein the plurality of user sets are generated using different user models; and

using ensemble learning to select one or more most relevant user sets from the plurality of user sets.

12. The computer readable medium of claim 11 wherein the ensemble learning is based, at least in part, on user behavior collected from user activity subsequent to previous recommendations.

13. The computer readable medium of claim 11 wherein the ensemble learning is further based, at least in part, on a naï

ve Bayesian network.

14. The computer readable medium of claim 11 wherein the naï

ve Bayesian network utilizes information concerning the particular retailer requesting recommendations, the page on which the recommendation is to be displayed and the one or more products.

15. The computer readable medium of claim 11 wherein an observed strength and reliability value for the selection of user sets is combined with a strength and reliability value predicted from the naï

ve Bayesian network.

16. An arrangement for improving product recommendations for a first user comprising:

a controller that receives a request for user set recommendations from any of a set of retailers where such request is related to one or more products, receives from a plurality of user sets from one or more automated user recommendation systems, wherein the plurality of user sets are generated using different user models, and uses ensemble learning to select one or more most relevant user sets from the plurality of user sets.

17. The arrangement of claim 16 wherein the ensemble learning is based, at least in part, on user behavior collected from user activity subsequent to previous recommendations.

18. The arrangement of claim 16 wherein the ensemble learning is further based, at least in part, on a naï

ve Bayesian network.

19. The arrangement of claim 16 wherein the naï

ve Bayesian network utilizes information concerning the particular retailer requesting recommendations, the page on which the recommendation is to be displayed and the one or more products.

20. The arrangement of claim 16 wherein an observed strength and reliability value for the selection of user sets is combined with a strength and reliability value predicted from the naï

ve Bayesian network.