INCREASING THE DIVERSITY OF ITEM RECOMMENDATIONS BY FILTERING

• US 20110173078A1
• Filed: 03/18/2011
• Published: 07/14/2011
• Est. Priority Date: 11/16/2007
• Status: Active Grant

• Current Assignee: Cory Hicks
• Original Assignee: Cory Hicks
• Inventors: Hicks, Cory
• Granted Patent: US 8,407,178 B2
• Time in Patent Office: Days
• US Class Current: 705/14.66
• CPC Class Codes: G06N 5/02 Knowledge representation; Symbolic representation
• CPC Class Codes: G06Q 30/0269 based on user profile or attribute

Abstract

A recommendation system increases the diversity of item recommendations provided to a target user by using item similarity data to filter an initial recommendation set. In one embodiment, selected items are filtered from the initial recommendation set based on similarity scores that represent degrees of similarity between particular items. The similarity scores may be generated based on an automated comparison of item attributes or content, or based on another measure of item similarity.

Claims

1. A computer-implemented method, comprising:

retrieving item preference data for a target user from computer storage, said item preference data reflective of item preferences of the target user;

generating a recommendation set using the retrieved item preference data, said recommendation set comprising a computer representation of items predicted to be of interest to the target user;

filtering the recommendation set, wherein filtering the recommendation set comprises filtering out a first item from the recommendation set based at least partly on a determination that the first item has at least a threshold degree of similarity to a second item in the recommendation set, said first and second items not being duplicates of each other, said filtering producing a filtered recommendation set that has a higher degree of item diversity than the recommendation set; and

outputting at least a portion of the the filtered recommendation set for presentation to the target user;

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

2. The method of claim 1, wherein outputting at least a portion of the filtered recommendation set comprises generating and transmitting a page that lists at least a portion of the filtered recommendation set, said page including a display element that is selectable by the target user to cause one or more filtered-out items, including said first item, to be displayed.

3. The method of claim 2, wherein the display element corresponds to the second item, and enables the target user to initiate the display of one or more filtered-out items that are similar to the second item.

4. The method of claim 1, wherein the method comprises comparing item attributes of the first and second items to determine whether said threshold degree of similarity is met.

5. The method of claim 1, wherein the method comprises (1) generating a similarity score that represents a degree of similarity between the first and second items, said similarity score reflecting a degree to which attributes of the first and second items are the same, and (2) using the similarity score to determine whether to filter out one of said first and second items.

6. The method of claim 1, wherein the first item is filtered out based additionally on a determination that the first item is ranked lower than the second item.

7. The method of claim 1, wherein the first item is filtered out based additionally on a determination that the first item and second item do not satisfy an exception condition that, if met, would permit the first and second items to be recommended in combination.

8. The method of claim 1, wherein filtering the recommendation set comprises using attribute-based item similarity scores to select items to filter from the recommendation set, each attribute-based item similarity score corresponding to a respective item pair.

9. The method of claim 1, wherein generating the recommendation set comprises using behavioral associations between items to select, for inclusion in the recommendation set, items that are behaviorally associated with items previously selected by the target user, said behavioral associations based on recorded behaviors of a plurality of users.

10. The method of claim 1, wherein the items are products represented in an electronic catalog.

11. A system for generating recommendations, the system comprising:

a data repository of item data, said item data including item attributes of a plurality of items;

an item comparison system configured to use the item data, including the item attributes, to generate similarity scores for particular item pairs, each similarity score representing a degree to which two particular items are similar to each other;

a data repository of item similarity data, said item similarity data identifying item pairs associated with similarity scores satisfying a selected threshold;

a recommendation engine configured to generate a recommendation set based on user preference data associated with a target user, said recommendation set comprising a computer representation of items predicted to be of interest to the target user; and

a similar items filter configured to use the data repository of item similarity data to identify, in said recommendation set, one or more item pairs whose respective similarity scores satisfy said threshold, said similar items filter additionally configured to filter out, from the recommendation set, one item of each such identified item pair, to thereby prevent certain items from being recommended to the target user in combination, said similar items filter comprising computer hardware.

12. The system of claim 11, wherein the similar items filter is capable of filtering out a first item from the recommendation set based at least partly on a determination that the first item has a threshold degree of similarity to a second item in the recommendation set, said first and second items not being duplicates of each other.

13. The system of claim 11, further comprising a server that is configured to generate, for display to the target user, a page that lists at least a portion of the filtered recommendation set, said page including one or more display elements for enabling the target user to initiate the display of one or more items filtered out from the recommendation set.

14. The system of claim 13, wherein the one or more display elements includes, for a first item in the filtered recommendation set, a display element for initiating the display of filtered-out items that are similar to said first item.

15. The system of claim 11, wherein the recommendation engine is operative to use behavior-based associations between items to generate the recommendation set.

16. Physical computer storage that stores executable code that directs a computing system to at least:

receive a recommendation set generated by a recommendation engine, said recommendation set comprising a computer representation of a plurality of items predicted to be of interest to a target user;

filter the recommendation set to increase a degree of item diversity of the recommendation set, said filtering comprising filtering out a first item from the recommendation set based at least partly on a determination that the first item has at least a threshold degree of similarity to a second item in the recommendation set, said first and second items not being duplicates of each other; and

output at least a portion of the filtered recommendation set for presentation to the target user.

17. The physical computer storage of claim 16, wherein the executable code directs the computing system to generate, for presentation to the target user, a display page that includes (1) a listing of items in the filtered recommendation set, and (2) a display element that is selectable to cause one or more filtered-out items to be displayed.

18. The physical computer storage of claim 16, wherein the executable code directs the computing system to compare item attributes of the first and second items to determine whether said threshold degree of similarity is met.

19. The physical computer storage of claim 16, wherein the executable code directs the computing system to (1) generate a similarity score that represents a degree of similarity between the first and second items, said similarity score reflecting a degree to which attributes of the first and second items are the same, and (2) use the similarity score to determine whether to filter out one of said first and second items.