CLUSTER-BASED ASSESSMENT OF USER INTERESTS

• US 20080243815A1
• Filed: 03/30/2007
• Published: 10/02/2008
• 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
• Granted Patent: US 7,689,457 B2
• Time in Patent Office: Days
• US Class Current: 707/5
• 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 30/0204 Market segmentation
• CPC Class Codes: G06Q 30/0255 based on user history
• CPC Class Codes: G06Q 30/0631 Item recommendations
• CPC Class Codes: G06Q 30/0641 Shopping interfaces

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 target user;

applying a clustering algorithm to the collection of items to subdivide the collection into multiple clusters of items, wherein the clustering algorithm generates said clusters based, at least in part, on calculated distances between the items;

selecting a subset of the items in the collection to use as recommendation source items based, at least in part, on one or more attributes of the clusters of items;

using the selected recommendation source items as inputs to a recommendation engine to generate a set of recommended items for the target user; and

outputting a representation of at least some of the recommended items to the target user.

2. The method of claim 1, wherein identifying the collection of items comprises using a recorded purchase history of the target user to identify items purchased by the target user.

3. The method of claim 1, wherein the collection of items comprises items selected from at least one of the following categories:

(1) items purchased by the target user, (2) items rated by the target user, (3) items rented by the target user, (4) items viewed by the target user, (5) items downloaded by the target user, (6) items subscribed-to by the target user.

4. The method of claim 1, wherein the items are represented in a hierarchical browse structure, and the method comprises calculating said distances between the items based, at least in part, on locations of the items in the hierarchical browse structure.

5. The method of claim 4, wherein the hierarchical browse structure includes multiple levels of browse nodes, and the method comprises calculating a distance between two items based on a degree to which the two items fall under common browse nodes, wherein different amounts of weight are given to different ones of said browse nodes.

6. The method of claim 1, wherein selecting the subset of items to use as recommendation sources comprises taking into consideration the number of items in each cluster.

7. The method of claim 6, wherein said numbers of items per cluster are taken into consideration such that an item falling in a relatively large cluster is more likely to be selected than an item falling in a relatively small cluster.

8. The method of claim 1, wherein selecting the subset of items to use as recommendation sources comprises excluding an item based on one or more attributes of a cluster in which the item falls.

9. The method of claim 1, wherein selecting the subset of items to use as recommendation sources comprises generating a score for a cluster, said score representing a likelihood that the cluster represents an interest of the target user.

10. The method of claim 1, wherein selecting the subset of items to use as recommendation sources comprises assessing whether a given cluster likely consists primarily of gift purchases made by the target user for another user.

11. The method of claim 1, wherein selecting the subset of items to use as recommendation sources comprises assessing whether a given cluster is close in space to items in which the user has explicitly indicated a lack of interest.

12. The method of claim 1, wherein the recommendation engine generates the set of recommended items by using a data repository of item similarity data to map the set of source items to a set of recommended items that are similar to the source items.

13. A computer system programmed to perform the method of claim 1.

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

15. A computer system, comprising:

a computer data repository that stores a collection of items associated with a user;

a clustering component that applies a clustering algorithm to the collection of items to divide the collection into multiple clusters of items;

a source selection component that selects items from said collection to use as recommendation sources, wherein the source selection component selects said items based, as least in part, on information regarding said clusters of items; and

a recommendation engine that uses the source items selected by the source selection component to generate personalized item recommendations for the user.

16. The computer system of claim 15, wherein the collection of items comprises items purchased by the user.

17. The computer system of claim 15, wherein the collection of items comprises items rated by the user.

18. The computer system of claim 15, wherein the items are arranged within a hierarchical browse structure having multiple levels of browse nodes, and the clustering component uses browse node ancestries of the items to calculate distances between the items.

19. The computer system of claim 18, wherein the clustering component gives different amounts of weight to different browse nodes in calculating said distances.

20. The computer system of claim 15, wherein the source selection component select the source items based, at least in part, on the number of items in each clusters.

21. The computer system of claim 15, wherein the source selection component is configured to generate a score for a cluster, and to use the score to determine whether to select any items from said cluster to use as sources.

22. The computer system of claim 21, wherein the score is dependent upon how the target user has rated one or more items in said cluster.

23. The computer system of claim 15, further comprising a user interface that enables the user to assign a tag to a selected one of said clusters, and to request item recommendations that are specific to said tag.

24. The computer system of claim 15, further comprising a user interface that enables the user to request item recommendations that are specific to a selected one of said clusters.

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

26. The computer system of claim 15, further comprising one or more processors that execute(s) the clustering component and the source selection component.

27. A computer-implemented method, comprising:

identifying a collection of items associated with a user, said collection represented in computer storage;

applying a clustering algorithm to the collection of items to subdivide the collection into a plurality of clusters of items, wherein the clustering algorithm generates said clusters based, at least in part, on calculated distances between the items;

for at least a one cluster of said plurality of clusters, assessing whether the cluster represents an interest of the user, said cluster comprising multiple items; and

storing a result of said assessment in computer storage.

28. The method of claim 27, wherein assessing whether the cluster represents an interest of the user comprises taking into consideration a size of the cluster.

29. The method of claim 27, wherein assessing whether the cluster represents an interest of the user comprises taking into consideration how the user rated one or more items in the cluster.

30. The method of claim 27, wherein the cluster comprises items purchased by the user, and wherein assessing whether the cluster represents an interest of the user comprises taking into consideration dates of purchase by the target user of items in the cluster.

31. The method of claim 27, further comprising determining, based at least in part on said assessment, whether to use items in said cluster as sources for generating recommendations for the user.

32. The method of claim 27, further comprising determining, based at least in part on said assessment, whether to use the cluster to define a category for presenting recommendations to the user.

33. A computer system programmed to perform the method of claim 27.

34. A computer-readable medium having stored thereon executable computer code that embodies the method of claim 27.