System for extrapolating item characteristics

• US 8,380,583 B1
• Filed: 12/23/2008
• Issued: 02/19/2013
• 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:
  US12/343,360
• Time in Patent Office: 1,519 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 extrapolating item characteristics to improve recommendations related to behaviorally-deficient items, the system comprising:

an item data repository comprising physical computer storage, the item data repository storing item data representative of a plurality of items in an electronic catalog, the plurality of items comprising first items having a characteristic and second items not known to have the characteristic;

an association mining module operative to programmatically analyze the item data to generate substitution associations between first ones of the plurality of items and behavioral associations between second ones of the plurality of items;

a characteristic extrapolation module operative to programmatically extrapolate the characteristic from at least some of the first items known to have the characteristic to at least some of the second items not known to have the characteristic based at least partly on a strength of the substitution associations, wherein the characteristic extrapolation module is further configured to extrapolate the characteristic by propagating the item characteristic through a directed graph, the directed graph comprising nodes and edges connecting the nodes, each of the nodes representing one of the plurality of items and being associated with a characteristic value that represents the characteristic, each of the edges representing one of the substitution associations, wherein the characteristic extrapolation module propagates the characteristic by at least updating the characteristic values of the nodes based at least in part on the strength of the substitution associations represented by the edges connecting the nodes; and

a recommendations module operative to generate for a user item recommendations that do not have the characteristic, the recommendations module configured to;

(1) access a selected item from user activity data corresponding to the user, the selected item having a deficiency of behavioral data,

(2) access one or more substitute items having substitution associations with the selected item,

(3) access one or more behavioral items having behavioral associations with the one or more substitute items, and

(4) select at least a portion of the behavioral items to recommend to the user that do not have the extrapolated characteristic.

2. The system of claim 1, wherein the characteristic is adultness.

3. The system of claim 1, wherein the characteristic extrapolation module propagates the characteristic through the directed graph based at least partly on a number of the substitution associations.

4. A system for extrapolating item characteristics from items considered to possess a characteristic to items not known to possess the characteristic, the system comprising:

an item data repository comprising physical computer storage, the item data repository storing item data representative of a plurality of items, the plurality of items comprising first items having a characteristic and second items not known to have the characteristic, the characteristic being an attribute other than an item association;

an association mining module operative to programmatically analyze the item data to generate associations between the plurality of items, the associations reflecting a degree of relatedness between the plurality of items;

a characteristic extrapolation module operative to programmatically extrapolate the characteristic from at least some of the first items known to have the characteristic to at least some of the second items not known to have the characteristic based at least partly on a strength of the associations between the plurality of items, wherein the characteristic extrapolation module is further configured to extrapolate the characteristic by propagating the item characteristic through a directed graph, the directed graph comprising nodes and edges connecting the nodes, each of the nodes representing one of the plurality of items and being associated with a characteristic value that represents the characteristic, each of the edges representing one of the associations, wherein the characteristic extrapolation module propagates the characteristic by at least updating the characteristic values of the nodes based at least in part on the strength of the associations represented by the edges connecting the nodes; and

a recommendations module operative to generate item recommendations for a user, the item recommendations being selected from at least some of the second items that de the characteristic extrapolation module considers not to have the characteristic.

5. The system of claim 4, wherein the characteristic extrapolation module propagates the characteristic through the directed graph based at least partly on a number of associations between the plurality of items.

6. The system of claim 4, wherein the recommendations module is further operative to filter out items possessing the characteristic from a set of candidate recommendations.

7. The system of claim 6, wherein the characteristic is adultness.

8. The system of claim 6, wherein the data representative of the plurality of items is stored in an electronic catalog.

9. A method of extrapolating item characteristics, the method comprising:

programmatically analyzing item data stored in an item data repository, the item data comprising data representative of a plurality of items in an electronic catalog, said analyzing comprising generating associations between the plurality of items, the associations reflecting a degree of relatedness between the plurality of items, wherein the plurality of items comprise first items considered to have a characteristic and second items not known to have the characteristic, the characteristic being an attribute other than an item association;

storing the associations in computer storage;

extrapolating the characteristic from at least some of the first items known to have the characteristic to at least some of the second items not known to have the characteristic based at least partly on a strength of the associations, said extrapolating the characteristic comprising;

propagating the item characteristic through a directed graph, the directed graph comprising nodes and edges connecting the nodes, each of the nodes representing one of the plurality of items and being associated with a characteristic value that represents the characteristic, each of the edges representing one of the associations, said propagating comprising updating the characteristic values of the nodes based at least in part on the strength of the associations represented by the edges connecting the nodes;

storing characteristic data comprising the characteristic value for each of the plurality of items; and

generating item recommendations for a user, the item recommendations being selected at least in part from the second items that do are not considered to have the characteristic as determined by said extrapolating;

wherein said analyzing, said storing the associations, said extrapolating, said storing characteristic data, and said generating are implemented by a computer system comprising computer hardware.

10. The method of claim 9, further comprising using the characteristic data to categorize at least some of the second items.

11. The method of claim 9, further comprising using the characteristic data to re-categorize items that have been mis-categorized.

12. The method of claim 9, further comprising using the characteristic data to select items to recommend to users.

13. The method of claim 9, programmatically analyzing the item data to generate associations between the plurality of items comprises creating substitution associations between the plurality of items.

14. The method of claim 9, further comprising using the characteristic data to improve recommendations related to behavior-deficient items.

15. The method of claim 9, further comprising using the characteristic data to prevent one or more types of items from being recommended to the user.