System for detecting probabilistic associations between items

• US 8,239,287 B1
• Filed: 01/15/2009
• Issued: 08/07/2012
• Est. Priority Date: 10/31/2008
• Status: Expired due to Fees

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Smith, Brent R., Whitman, Ronald, Chanda, Gaurav
• Primary Examiner(s): Smith, Jeffrey A
• Assistant Examiner(s): Georgalas, Anne
• Application Number:
  US12/354,233
• Time in Patent Office: 1,300 Days
• Field of Search: 705/26.27, 705/26.1, 705/26.7, 705/27.1
• US Class Current: 705/26.7
• CPC Class Codes: G06Q 30/0241 Advertisements

Abstract

A computer system for detecting associations between items may include a probabilistic analysis component that accesses user data having information about item selections by a plurality of users. The probabilistic analysis component may programmatically generate associations between certain items by determining a first number of users who selected the items and by estimating a probability that a second number of users would have selected the items due to random chance. The probabilistic analysis component may estimate this probability based at least partly on a number of item selections made by one or more of the users.

Claims

1. A system for using probabilistically-generated item associations to provide personalized item recommendations, the system comprising:

a computer system comprising one or more processors, the computer system comprising;

a probabilistic analysis component executed by the one or more processors, the probabilistic analysis component configured to;

access item data from a data repository, the item data comprising information regarding a plurality of items, the data repository comprising physical computer storage;

access user data comprising information about item selections by a plurality of users;

programmatically generate an association between a first item and a second item of the item data based at least in part on the user data by at least;

determining, from the user data, a first probability that first users selected both the first and second items,

estimating, from the user data, a second probability that second users who selected the first item would have selected the second item independent of selecting the first item, said estimating comprising

     1) determining a number of item selections other than the selection of the first item made by each of the second users and

     2) using the number of item selections to estimate a probability that one of the other item selections was the second item, and

deriving an association score for the first and second items based at least in part on a difference between the first probability and the second probability, the association score reflecting a value of the association between the first and second items; and

a recommendation module executed by the one or more processors, the recommendation module configured to use the association to generate one or more item recommendations for a target user.

2. The system of claim 1, wherein the items comprise items in an electronic catalog.

3. The system of claim 2, wherein the item selections comprise one or more of purchases, views, downloads, rentals, and clicks.

4. The system of claim 1, wherein the probabilistic analysis component is further configured to programmatically generate the association by transforming at least part of the item data into the association.

5. The system of claim 1, wherein the computer system comprises multiple physical computers.

6. A computer-implemented method of creating associations between physical items represented in a data repository, the method comprising:

by a computer system comprising computer hardware;

accessing item data from a data repository, the item data comprising information representing a plurality of physical items, the data repository comprising physical computer storage;

accessing user data comprising information about item selections by a plurality of users; and

programmatically generating an association between a first item and a second item of the item data based at least in part on the user data, said generating comprising;

determining, from the user data, a first probability that first users selected both the first and second items,

estimating, from the user data, a second probability that second users who selected the first item would have selected the second item independent of selecting the first item by at least

     1) determining a number of item selections other than the selection of the first item made by each of the second users and

     2) using the number of item selections to estimate a probability that one of the other item selections was the second item, and

deriving an association score for the first and second items based at least in part on the first and second probabilities, the association score reflecting a value of the association between the first and second items.

7. The method of claim 6, wherein the items comprise items in an electronic catalog.

8. The method of claim 7, wherein the item selections comprise one or more of purchases, views, downloads, rentals, and clicks.

9. The method of claim 6, further comprising using the association between the first item and the second item to recommend one or more items to a user.

10. The method of claim 6, wherein said programmatically generating the association comprises transforming at least part of the item data into the association.

11. The method of claim 6, wherein said determining a number of item selections comprises calculating an average number of the item selections other than the selection of the first item made by each of the second users.

12. The method of claim 6, wherein the computer system comprises one or more processors.

13. The method of claim 6, wherein said deriving an association score for the first and second items based at least in part on the first and second probabilities comprises deriving an association score for the first and second items based at least in part on a difference between the first probability and the second probability.

14. A computer system for detecting associations between items represented in a data repository, the system comprising:

a probabilistic analysis component comprising computer hardware, the probabilistic analysis component configured to;

access user data comprising information about item selections by a plurality of users; and

programmatically generate associations between first and second items based at least in part on the user data by at least;

determining, from the user data, a first probability that first users selected both the first and second items,

estimating, from the user data, a second probability that second users who selected the first item would have selected the second item based at least partly on a number of item selections other than the selection of the first item made by each of the second users,

deriving an association score for the first and second items based at least in part on the first and second probabilities, the association score reflecting a degree of the association between the first and second items, and

store the association in a data repository.

15. The system of claim 14, wherein the items comprise items in an electronic catalog.

16. The system of claim 15, wherein the item selections comprise one or more of purchases, views, downloads, rentals, and clicks.

17. The system of claim 14, further comprising using the association between the first and second items to recommend one or more items to a user.

18. The system of claim 14, wherein the probabilistic analysis component is further configured to programmatically generate associations by transforming at least part of the item data into the association.

19. The system of claim 14, wherein the probabilistic analysis component is further configured to approximate the estimated second probability by calculating an average number of item selections.

20. The system of claim 14, wherein the computer system comprises multiple physical computers.

21. The system of claim 14, wherein the probabilistic analysis component is further configured to estimate the second probability by at least using the number of item selections to estimate a probability that one of the other item selections was the second item.

22. The system of claim 14, wherein said deriving an association score for the first and second items based at least in part on the first and second probabilities comprises deriving an association score for the first and second items based at least in part on a difference between the first probability and the second probability.