Cluster-based management of collections of items

• US 7,743,059 B2
• Filed: 03/30/2007
• Issued: 06/22/2010
• Est. Priority Date: 03/30/2007
• Status: Expired due to Fees

• 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): Ali; Mohammad
• Assistant Examiner(s): Darno; Patrick A
• Application Number:
  US11/694,815
• Publication Number:
  US 20080243817A1
• Time in Patent Office: 1,180 Days
• Field of Search: None
• US Class Current: 707/737
• 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 30/0631 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 of generating personalized recommendations of items, the method comprising:

maintaining an item collection of a user in computer storage, said item collection being a computer representation of a plurality of items selected by the user from an electronic catalog of items, wherein the items are represented in a hierarchical browse structure of said electronic catalog, said hierarchical browse structure comprising multiple levels of browse nodes;

applying a clustering algorithm to the item collection to subdivide the collection into multiple clusters of said items, said clusters generated based, at least in part, on calculated distances between the items in the hierarchical browse structure, said distances being dependent upon the browse nodes to which particular items are assigned;

outputting a visual representation of the multiple clusters for presentation to the user via a user interface that enables the user to rate specific clusters of items, said visual representation enabling the user to identify particular items included in each of said clusters;

receiving an indication of a cluster rating specified by the user via said user interface, said cluster rating corresponding to a cluster selected by the user from said visual representation, and representing a collective rating by said user of multiple items in said cluster;

generating personalized item recommendations for the user by execution of code modules by a computer system, wherein generating the personalized item recommendations comprises taking the cluster rating into consideration in selecting items from the item collection to use as recommendation sources, and by identifying, for recommendation to the user, additional items that are related to said recommendation source items, such that the personalized item recommendations are based on a selected subset of the item collection; and

outputting a representation of the personalized item recommendations for presentation to the user.

2. The method of claim 1, wherein the items represented in the item collection include or consist of items purchased by the user.

3. The method of claim 1, wherein the items represented in the item collection include or consist of items rated by the user.

4. The method of claim 1, wherein generating the personalized item recommendations comprises using a pre-generated dataset of item-to-item association mappings to look up additional items that are related to the recommendation source items.

5. The method of claim 1, wherein the method comprises using said cluster rating as a default item rating for items in the cluster that have not been rated by the user, while preserving item ratings of items in the cluster that have been rated by the user.

6. The method of claim 1, wherein generating the personalized item recommendations additionally comprises using at least the cluster rating to determine how much weight to give to individual recommendation source items in generating said personalized item recommendations.

7. The method of claim 1, wherein generating the personalized item recommendations further comprises taking item-specific ratings assigned to particular items by the user into consideration in selecting said items from the item collection to use as recommendation sources.

8. The method of claim 1, wherein the cluster rating is specified on a scale of one to five.

9. The method of claim 1, wherein the indication of the cluster rating is an indication by the user of a least one of the following:

(a) the user lacks interest in the cluster, and (b) the cluster should not be used to generate recommendations.

10. The method of claim 1, wherein the method comprises using at least the cluster rating to select multiple items from the collection to use as recommendation sources for generating said item recommendations, such that less than all of the items represented in the user'"'"'s item collection are used as recommendation sources for generating the personalized item recommendations.

11. The method of claim 1, further comprising providing the user an option, via said user interface, to request and obtain item recommendation that are specific to a particular one of said clusters.

12. The method of claim 1, further comprising receiving user input that specifies assignments of category names to particular ones of said clusters, and using said category names to arrange the personalized item recommendations for presentation to the user.

13. A computer system programmed to perform the method of claim 1, said computer system comprising one or more physical computers.

14. A computer-readable medium having stored thereon executable computer code that embodies the method of claim 1.

32. The method of claim 1, wherein the item collection is a representation of a plurality of catalog items selected by the user for performing at least one of the following types of actions:

purchase, rent, view, download, add to a shopping cart, add to a wish list.

33. The method of claim 1, wherein each of the clusters corresponds to a respective area of interest associated with the user.

34. The method of claim 33, wherein the computer-implemented method comprises using at least the cluster rating to select the recommendation source items exclusively from a particular cluster corresponding to a particular area of interest associated with the user, such that items included in said collection but falling outside said area of interest are not used as recommendation sources.

35. The method of claim 1, wherein the method comprises selecting the recommendation source items exclusively from one of said multiple clusters.

36. The method of claim 1, wherein applying the clustering algorithm comprises calculating a distance between a first item and a second item such that said distance depends on how many browse nodes are shared by the first and second items.

37. The method of claim 1, wherein the items are products that are available for purchase via said electronic catalog.

38. The method of claim 1, further comprising reassigning an item from a first one of said clusters to a second one of said clusters in response to a drag-and-drop operation performed via said user interface.

39. The method of claim 1, wherein the computer system comprises a plurality of physical servers of a web site that hosts said electronic catalog.

15. A computer system, comprising:

a computer data repository that stores a collection of items associated with a user, said collection of items comprising a computer representation of a plurality of items selected by the user from an electronic catalog of items, wherein the items are arranged in the electronic catalog according to a hierarchical browse structure that has multiple levels of browse nodes;

a clustering component that is operative to apply a clustering algorithm to the collection of items to divide the collection into multiple clusters of items, said clustering component operative to generate the clusters based at least partly on distances between particular items in the hierarchical browse structure, said distances being dependent upon the browse nodes to which the items are assigned;

a collection management interface that includes functionality for the user to view the clusters of items, and to visualize assignments of particular items to particular clusters, said cluster management interface additionally providing functionality for the user to assign a cluster rating to a cluster to thereby collectively rate a plurality of items included in said cluster; and

a recommendation system that is operative to use at least said cluster rating to select, from said collection of items, a subset of items to use as recommendation sources, and to use said subset of items to generate personalized item recommendations for the user, such that the personalized item recommendations are based on less than all of the items of the item collection, said recommendation system comprising computer hardware.

16. The computer system of claim 15, wherein the items represented in the item collection include or consist of items purchased by the user.

17. The computer system of claim 15, wherein the items represented in the item collection include or consist of items rated by the user.

18. The computer system of claim 15, wherein the hierarchical browse structure includes browse nodes that correspond to particular item categories and subcategories, and wherein the clustering component uses browse node ancestries of the items to calculate said distances between the items.

19. The computer system of claim 15, wherein the recommendation system is operative to generate the personalized item recommendations at least in part by using pre-generated item relationship data to identify additional items that are related to the recommendation source items.

20. The computer system of claim 15, wherein the recommendation system is operative to use at least the cluster rating to determine how much weight to give to particular recommendation source items in generating said item recommendations.

21. The computer system of claim 15, wherein the collection management interface additionally provides functionality for the user to assign tags to particular clusters of items, and the recommendation system takes said tags into consideration in generating the personalized item recommendations.

22. The computer system of claim 15, wherein the collection management interface provides functionality for the user to request and obtain item recommendations that are specific to a particular one of said clusters.

23. The computer system of claim 15, wherein the collection management interface provides functionality for the user to assign a category name to a cluster, and the recommendation system uses said category name to present associated item recommendations to the user.

24. The computer system of claim 15, further comprising a recommendation categorization component that uses the clusters generated by the clustering component to subdivide the personalized item recommendations into multiple cluster-based categories for presentation to the user.

25. The computer system of claim 15, wherein the item collection is a representation of a plurality of catalog items selected by the user for performing at least one of the following types of actions:

purchase, rent, view, download, add to a shopping cart, add to a wish list.

26. The computer system of claim 15, wherein the clustering component subdivides the collection of items into clusters such that each cluster corresponds to a respective area of interest associated with the user.

27. The computer system of claim 26, wherein the recommendation system is operative to select said subset solely from a particular cluster corresponding to a particular area of interest, such that items included in said collection but falling outside said area of interest are not used as recommendation sources.

28. The computer system of claim 15, wherein the subset of items consists of a plurality of items in one of said multiple clusters.

29. The computer system of claim 15, wherein the items are products that are available for purchase via said electronic catalog.

30. The computer system of claim 15, further comprising at least one web server machine that provides interactive user access to said collection management interface.

31. The computer system of claim 15, further comprising an interactive system that includes said computer data repository, said clustering component, said collection management interface, and said recommendation system, said interactive system comprising one or more physical servers.

40. The computer system of claim 15, wherein the collection management interface includes functionality for the user to reassign an item to a different cluster via a drag-and-drop operation.

41. The computer system of claim 15, wherein the cluster rating is on a scale of one to five.