System for extrapolating item characteristics

• US 8,751,333 B1
• Filed: 02/14/2013
• Issued: 06/10/2014
• Est. Priority Date: 12/23/2008
• Status: Active Grant

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Chanda, Gaurav
• Primary Examiner(s): Shaawat, Mussa A
• Application Number:
  US13/767,761
• Time in Patent Office: 481 Days
• Field of Search: None
• US Class Current: 705/26.7
• CPC Class Codes: G06Q 30/0255 based on user history

Abstract

A system is provided that extrapolates item characteristics from items considered to possess a characteristic to items not known to possess the characteristic. The system may include an item data repository that stores data representing physical items. These items can include first items having a characteristic and second items not known to have the characteristic. A characteristic extrapolation module can extrapolate the characteristic from at least some of the first items to at least some of the second items based at least in part on the strength of associations between the plurality of items. A recommendations module may provide item recommendations based at least partly on the characteristic of the items.

Claims

1. A system for preventing inappropriate items from being recommended to a user, the system comprising:

an association mining module configured to programmatically analyze item data regarding a plurality of items to generate substitution associations between the plurality of items;

a characteristic extrapolation module comprising computer hardware, the characteristic extrapolation module configured to extrapolate a characteristic from a first set of the plurality of items known to have the characteristic to a portion of a second set of the plurality of items not known to have the characteristic based at least partly on a strength of the substitution associations, the characteristic representing a feature of the first set of items that is stored in descriptions associated with the first set of items in an electronic catalog but not the second items, wherein the characteristic is adultness and is not an item association; and

a recommendations module configured to generate item recommendations, the item recommendations being selected from at least some of the second set of the plurality of items that are not considered by the characteristic extrapolation module to have the characteristic.

2. The system of claim 1, wherein the recommendations module is further configured to select the item recommendations based on user activity data of a user.

3. The system of claim 2, wherein the recommendations module is further configured to provide the item recommendations to the user.

4. The system of claim 1, wherein the characteristic extrapolation module is further configured to extrapolate the characteristic by using the substitution associations to categorize the second set of the plurality of items as adult items in an electronic catalog.

5. The system of claim 1, wherein the characteristic extrapolation module is further configured to use the substitution associations to re-categorize adult items that have been mis-categorized as non-adult items.

6. The system of claim 1, wherein the characteristic extrapolation module is further configured to extrapolate the characteristic by propagating the item characteristic through a graph comprising nodes, each node representing a selected one of the plurality of items.

7. The system of claim 6, wherein the graph is an undirected graph.

8. A method of preventing inappropriate items from being recommended to a user, the method comprising:

generating substitution associations between first ones of a plurality of items;

generating behavioral associations between second ones of the plurality of items;

extrapolating a characteristic from a first set of the plurality of items known to have the characteristic to a first portion of a second set of the plurality of items not known to have the characteristic based at least partly on a strength of the substitution associations, the characteristic representing a feature of the first set of items that is stored in descriptions associated with the first set of items in an electronic catalog but not the second items, wherein the characteristic is adultness and is not an item association; and

selecting a second portion of the second set of the plurality of items that do not have the characteristic as recommendations;

wherein at least said extrapolating is performed by a computer system comprising computer hardware.

9. The method of claim 8, further comprising selecting the portion of the second set of the plurality of items as the recommendations based on user activity data of a user.

10. The method of claim 9, further comprising providing the recommendations to the user.

11. The method of claim 8, wherein said extrapolating further comprises using the substitution associations to categorize the second set of the plurality of items as adult items in an electronic catalog.

12. The method of claim 8, wherein said extrapolating further comprises using the substitution associations to re-categorize adult items that have been mis-categorized as non-adult items.

13. The method of claim 8, wherein said extrapolating further comprises representing one or both of the substitution associations and the behavioral associations in a graph.

14. The method of claim 13, wherein the graph is an undirected graph.

15. A method of preventing inappropriate items from being recommended to a user, the method comprising:

identifying a first item in an electronic catalog of items, the first item comprising a first description stored in the electronic catalog, the first description indicating that the first item possesses a characteristic, wherein the characteristic does not refer to an item association with another one of the items in the electronic catalog;

identifying a second item in the electronic catalog, the second item comprising a second description stored in the electronic catalog, the second description not including the characteristic;

evaluating the first and second items to determine a quantity representative of a degree of association between the first and second items based on one or more criteria other than the characteristic of the first item;

inferring that the second item has the characteristic based on the quantity representative of the degree of relationship between the first and second items, rendering the second item potentially inappropriate as a recommendation;

recommending a set of items from the electronic catalog of items to a user while excluding the second item from the recommended set of items due at least in part to said inferring that the second item has the characteristic;

wherein at least said inferring is performed by a computer system comprising computer hardware.

16. The method of claim 15, wherein the characteristic is adultness.

17. The method of claim 15, wherein the characteristic is represented by one or both of a numerical value and a textual description in the first description stored in the electronic catalog.

18. The method of claim 17, wherein the second description of the second item does not include the numerical value or the textual description of the first description of the first item.

19. The method of claim 15, wherein said evaluating comprises comparing one or both of attributes and a category of the first and second items.

20. The method of claim 15, wherein said evaluating comprises determining an extent to which the first and second items were both selected by users in an electronic catalog browsing session.

21. The method of claim 15, further comprising programmatically re-categorizing the second item in the electronic catalog based on said inferring.