Method, system, and medium for cluster-based categorization and presentation of item recommendations

• US 7,966,225 B2
• Filed: 03/30/2007
• Issued: 06/21/2011
• Est. Priority Date: 03/30/2007
• Status: Active Grant

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Ionkov, George M., Chan, James D., Chakrabarti, Kushal
• Primary Examiner(s): Garg; Yogesh C
• Assistant Examiner(s): ZIMMERMAN, MATTHEW E
• Application Number:
  US11/694,821
• Publication Number:
  US 20080243638A1
• Time in Patent Office: 1,544 Days
• Field of Search: 705/26.7, 707/737, 707/751
• US Class Current: 705/26.7
• CPC Class Codes: G06F 16/2457 with adaptation to user needs
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0253 During e-commerce, i.e. online transactions
• CPC Class Codes: G06Q 30/0631 Item recommendations

Abstract

Computer-implemented processes are disclosed for clustering items and improving the utility of item recommendations. One process involves applying a clustering algorithm to a user'"'"'s collection of items. Information about the resulting clusters is then used to select items to use as recommendation sources. Another process involves displaying the clusters of items to the user via a collection management interface that enables the user to attach cluster-level metadata, such as by rating or tagging entire clusters of items. The resulting metadata may be used to improve the recommendations generated by a recommendation engine. Another process involves forming clusters of items in which a user has indicated a lack of interest, and using these clusters to filter the output of a recommendation engine. Yet another process involves applying a clustering algorithm to the output of a recommendation engine to arrange the recommended items into cluster-based categories for presentation to the user.

Claims

1. A computer-implemented method, comprising:

identifying a collection of items associated with a user, said collection based on item selection actions of the user;

dividing the collection of items into clusters based, at least in part, upon calculated distances between items;

generating a set of recommended items for the user based, at least in part, on an analysis of the clusters using a computer;

dividing at least a portion of the set of recommended items into multiple clusters of recommended items, wherein the clusters of recommended items are generated based, at least in part, on calculated item distances;

assigning a category name to each cluster of recommended items; and

exposing each cluster of recommended items to the user in association with its respective assigned category name to facilitate selection of particular recommended items for consumption.

2. The method of claim 1, wherein the method is performed so as to guarantee that the clusters of recommended items are mutually exclusive.

3. The method of claim 1, wherein generating the set of recommended items comprises passing a list of source items to a recommendation engine, and receiving from the recommendation engine a corresponding list of said recommended items.

4. The method of claim 1, wherein dividing at least a portion of the set of recommended items into multiple clusters comprises applying a clustering algorithm to the set of recommended items, wherein the clustering algorithm uses calculated distances between the recommended items to form the clusters.

5. The method of claim 1, wherein dividing at least a portion of the set of recommended items into multiple clusters comprises dividing the set of recommended items into clusters that correspond to previously-identified interests of the user.

6. The method of claim 5, wherein the clusters of recommended items are formed at least partly by matching particular recommended items to particular ones of said clusters formed by dividing the collection of items.

7. The method of claim 6, wherein at least one of the clusters formed by dividing the collection has a name assigned by the user, and the step of assigning a category name comprises assigning the user-assigned name to a corresponding cluster of recommended items.

8. The method of claim 1, wherein the item distances used to divide at least a portion of the set of recommended items into multiple clusters include distances between particular recommended items and centers of pre-existing clusters.

9. The method of claim 1, wherein the items are represented in a hierarchical browse structure, and wherein dividing at least a portion of the set of recommended items comprises using locations of the recommended items in the hierarchical browse structure to calculate said item distances.

10. The method of claim 9, wherein using locations of the recommended items in the hierarchical browse structure to calculate said item distances comprises giving different amounts of weight to different browse nodes of the hierarchical browse structure.

11. The method of claim 1, wherein assigning a category name to each cluster of recommended items comprises selecting a category name for a first cluster based, at least in part, on attributes of the recommended items in the first cluster.

12. The method of claim 11, wherein the attributes include browse nodes to which the recommended items are assigned.

13. The method of claim 11, wherein the attributes include keywords associated with particular recommended items.

14. The method of claim 1, wherein assigning a category name to each cluster of recommended items comprises selecting a category name corresponding to a browse node under which all of the items in the respective cluster fall.

15. The method of claim 1, wherein exposing each cluster of recommended items comprises, by a server, exposing the clusters of recommended items via a browse cloud interface that displays the category names.

16. The method of claim 1, wherein exposing each cluster of recommended items comprises generating a display page in which each category name is selectable to view a representation of the items falling in the respective cluster.

17. The method of claim 16, wherein each of the category names is displayed with a font size that is proportional to the number of items falling in the respective cluster.

18. The method of claim 1, wherein generating the set of recommended items comprises:

selecting source items from the collection based, at least in part, on information obtained from dividing the collection into clusters; and

identifying items that are related to the selected source items.

19. The method of claim 1 wherein dividing the collection into clusters comprises forming a not-interested item cluster comprising items in which the user has indicated a lack of interest, and wherein generating the set of recommended items comprises filtering out, from a preliminary set of item recommendations, items that are associated with said not-interested item cluster.

20. A computer system, comprising:

a computer data repository that stores an item collection associated with a user, said item collection based on item selection actions of the user;

a first clustering component configured to divide the item collection into multiple clusters of collection items based, at least in part, on calculated item distances;

a recommendation system configured to generate personalized item recommendations for the user based, at least in part, on the clusters of collection items;

a second clustering component configured to use distance calculations to divide at least a portion of the personalized item recommendations into multiple clusters of recommended items, and to assign category names to said clusters of recommended items; and

a server configured to expose the clusters of recommended items to the user in association with said category names to facilitate selection of particular recommended items for consumption.

21. The computer system of claim 20, wherein the second clustering component uses a clustering process that guarantees that the clusters of recommended items are mutually exclusive.

22. The computer system of claim 20, wherein the second clustering component generates the clusters of recommended items, at least in part, by applying a clustering algorithm to the personalized item recommendations.

23. The computer system of claim 20, wherein the second clustering component generates the clusters of recommended items, at least in part, by matching individual recommended items to previously-identified interests of the user to generate interest-specific clusters of recommended items.

24. The computer system of claim 23, wherein the second clustering component identifies said interests, at least in part, based on the clusters of collection items.

25. The computer system of claim 24, wherein the server provides a user interface that enables the user to manage the item collection at a cluster level, including assigning a tag to multiple items at a time by tagging a selected cluster of collection items, said user interface also providing functionality for requesting item recommendations that are specific to said tag.

26. The computer system of claim 24, further comprising a user interface that enables the user to request and obtain recommendations that are specific to a cluster of items of said item collection.

27. The computer system of claim 20, wherein the second clustering component generates the clusters of recommended items, at least in part, by matching particular recommended items to particular clusters of collection items.

28. The computer system of claim 20, wherein the items are represented in a hierarchical browse structure, and the second clustering component uses locations of the recommended items in the hierarchical browse structure to generate said distance calculations.

29. The computer system of claim 28, wherein the second clustering component gives different amounts of weight to different browse nodes of the hierarchical browse structure when generating said distance calculations.

30. The computer system of claim 20, wherein the second clustering component is configured to assign a category name to a cluster of recommended items based, at least in part, on attributes of the recommended items in the cluster of recommended items.

31. The computer system of claim 30, wherein the attributes include browse nodes associated with particular recommended items.

32. The computer system of claim 30, wherein the attributes include keywords associated with particular recommended items.

33. The computer system of claim 20, wherein the server exposes the clusters of recommended items via a browse cloud interface that displays the category names.

34. The computer system of claim 20, wherein the server generates a display page in which each category name is selectable to view a representation of the recommended items falling in the respective cluster.

35. Non-transitory computer storage that stores executable code that directs a computer system to perform a process that comprises:

identifying a collection of items associated with a user, said collection based on item selection actions of the user;

dividing the collection of items into clusters based, at least in part, upon calculated distances between items;

generating a set of recommended items for the user based, at least in part, on an analysis of the clusters;

dividing at least a portion of the set of recommended items into multiple clusters of recommended items, wherein the clusters of recommended items are generated based, at least in part, on calculated item distances;

assigning a category name to each cluster of recommended items; and

exposing each cluster of recommended items to the user in association with its respective assigned category name to facilitate selection of particular recommended items for consumption.