Extrapolation of behavior-based associations to behavior-deficient items

• US 8,032,425 B2
• Filed: 06/16/2006
• Issued: 10/04/2011
• Est. Priority Date: 06/16/2006
• Status: Active Grant

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Yi, Jin Y.
• Primary Examiner(s): SHAAWAT, MUSSA A
• Application Number:
  US11/424,730
• Publication Number:
  US 20080004989A1
• Time in Patent Office: 1,936 Days
• Field of Search: None
• US Class Current: 705/26.7
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0601 Electronic shopping [e-shopping]
• CPC Class Codes: G06Q 30/0629 for generating comparisons
• CPC Class Codes: G06Q 30/0631 Item recommendations

Abstract

Behavior-based associations are extrapolated to items for which the quantity of collected user activity data is insufficient to create meaningful or reliable behavior-based associations (“behavior-deficient” items). The behavior-based associations are extrapolated based on content-based associations, or another type of “substitutability” association, between the behavior-deficient items and other items. The items can be any type of item for which user behaviors (e.g., purchases, accesses, downloads, etc.) can be monitored and analyzed to detect behavior-based associations, and for which item content or other available information can be used to assess item substitutability. For example, the items can be products represented in an electronic catalog, web pages or other documents accessible on a network, or web sites.

Claims

1. A computer-implemented method of detecting associations between items represented in an electronic catalog, the method comprising:

maintaining item selection histories of each of a plurality of users of the electronic catalog, said item selection histories identifying particular items selected by particular users;

detecting a behavior-based association between two items represented in the electronic catalog, item A and item B, at least in part by programmatically analyzing said item selection histories, wherein detecting the behavior-based association comprises generating a first data value that represents a strength of the behavior-based association between items A and B;

detecting a substitutability association between item A and an additional item, item C, represented in the electronic catalog, wherein detecting the substitutability association comprises generating a second data value that represents a strength of the substitutability association between items A and C;

creating an association between item B and item C based at least partly on the detected behavior-based association between items A and B in combination with the detected substitutability association between items A and C, wherein creating said association comprises generating a third data value that represents a strength of the association between items B and C, such that the third data value is dependent on the first and second data values; and

using the created association between item B and item C as a basis for selecting an item to recommend to one or more users;

said method performed in its entirety by a computer system that comprises one or more computers.

2. The method of claim 1, wherein the substitutability association is detected, at least in part, by analyzing attributes of items A and C.

3. The method of claim 1, wherein the substitutability association is detected, at least in part, by analyzing item viewing histories of users of the electronic catalog.

4. The method of claim 3, wherein the behavior-based association is detected based on purchase histories of users.

5. The method of claim 1, wherein item C is a behavior-deficient item.

6. The method of claim 5, further comprising phasing out or terminating the created association between items B and C in response to the collection of sufficient item selection history data for creating behavioral-based associations between item C and other items.

7. The method of claim 1, wherein item B and item C are not substitutable.

8. The method of claim 1, wherein the association created between items B and C is a directional association from item C to item B.

15. The method of claim 1, wherein creating the association between item B and item C is performed at least partly in response to a programmatic determination that item C is a behavior-deficient item.

16. The method of claim 1, wherein using the created association as a basis for selecting an item to recommend comprises using the association created between items B and C as a basis for supplementing an electronic catalog page for item C with an indication that item B is related to item C.

17. The method of claim 1, wherein the method comprises using the created association as a basis to select item C to recommend to a user based on a selection by the user of item B.

9. A non-transitory computer-readable storage medium or system having stored thereon executable instructions that, when executed, cause a computer system to perform a method that comprises:

storing item selection histories of each of a plurality of users of an electronic catalog, said item selection histories identifying particular items selected by particular users;

detecting a behavior-based association between two items represented in the electronic catalog, item A and item B, at least in part by analyzing said item selection histories, wherein detecting the behavior-based association comprises generating a first data value that represents a strength of the behavior-based association between items A and B;

detecting a substitutability association between item A and an additional item, item C, represented in the electronic catalog, wherein detecting the substitutability association comprises generating a second data value that represents a strength of the substitutability association between items A and C;

creating, in computer storage, an association between item B and item C based at least partly on the detected behavior-based association between items A and B in combination with the detected substitutability association between items A and C, wherein creating said association comprises generating a third data value that represents a strength of the association between items B and C, such that the third data value is dependent on the first and second data values; and

selecting an item to recommend based at least partly on the association created between item B and item C.

18. The non-transitory computer storage medium or system of claim 9, wherein the substitutability association is detected based at least partly on an automated comparison of item attributes of items A and C.

10. A system for detecting associations between items represented in an electronic catalog of items, the system comprising:

a computer system comprising one or more computers, said computer system configured to implement;

an item data repository that stores catalog descriptions of items represented in the electronic catalog;

a substitutability association mining component that detects substitutability associations between particular items represented in the electronic catalog, at least in part, by analyzing said catalog descriptions;

a data repository that stores item selection histories of a plurality of users of the electronic catalog, said item selection histories identifying particular items selected by particular users;

a behavior-based association mining component that detects behavior-based associations between particular items, at least in part, by analyzing the item selection histories of the plurality of users; and

an extrapolation component that detects additional associations between items by combining the detected substitutability associations with the detected behavior-based associations, said extrapolation component configured to extrapolate a behavior-based association between items A and B to item C by a process that comprises using at least a first data value that represents a strength of the behavior-based association between items A and B, in combination with a second data value that represents a strength of a substitutability association between items A and C, to calculate a third data value that represents a strength of an association between items B and C.

11. The system of claim 10, wherein the extrapolation component is capable of creating an association between two items between which no behavior-based association and no content-based association exists.

12. The system of claim 10, wherein the substitutability association mining component additionally uses item viewing histories of users to detect said substitutability associations.

13. The system of claim 10, wherein the item selection histories are item purchase histories.

14. The system of claim 10, further comprising a recommendation system that uses the associations detected by the extrapolation component to select items to recommend to users.