Methods and system of associating reviewable attributes with items

• US 8,645,295 B1
• Filed: 07/27/2009
• Issued: 02/04/2014
• Est. Priority Date: 07/27/2009
• Status: Active Grant

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Dillard, Logan Luyet
• Primary Examiner(s): Chaki, Kakali
• Assistant Examiner(s): Tran, Mai T
• Application Number:
  US12/510,212
• Time in Patent Office: 1,653 Days
• Field of Search: 706/20, 705/7.29
• US Class Current: 706/20
• CPC Class Codes: G06N 3/08 Learning methods
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0278 Product appraisal
• CPC Class Codes: G06Q 30/0282 Rating or review of business operators or products

Abstract

Users are enabled to provide structured ratings for various attributes of items or other such content in an electronic environment. Users are able to rate existing attributes associated with an item, or new attributes that the users want to associate with the item. In addition to allowing users to provide a rating for each attribute, users can be prompted to include information relating to these attributes in reviews for the respective item(s). Attributes can be automatically applied to various items using a process that determines aspects of items that are indicative of each attribute being relevant, and automatically applies the attributes to items having at least some of those or similar aspects. Various models and algorithms are described for providing such functionality.

Claims

1. A computer-implemented method of associating reviewable attributes with items, comprising:

under control of one or more computer systems configured with executable instructions,

for a first item in a plurality of items in a catalog, identifying a reviewable attribute of the first item as being configured to receive a review from a user, the reviewable attribute including a placeholder element for receiving attribute review text associated with the reviewable attribute of the first item;

storing the attribute review text associated with the reviewable attribute of the first item for subsequent display with the first item and information relating to the first item;

analyzing the reviewable attribute associated with the first item and information relating to the first item;

creating a model for the reviewable attribute using the information for the first item associated with the reviewable attribute, the model configured to associate the reviewable attribute with at least a second item such that the second item is configured with a new reviewable attribute placeholder element;

applying the model to at least a subset of the plurality of items to determine whether the second item in the subset should be associated with the reviewable attribute;

when it is determined that the reviewable attribute should be associated with the second item, storing the association of the reviewable attribute to the second item to enable the new reviewable attribute placeholder element; and

in response to a request from the user to review the second item, providing for display at least the new reviewable attribute placeholder element based at least in part on the stored association, the new reviewable attribute placeholder element configured to enable the user to supply a review for the reviewable attribute.

2. The computer-implemented method of claim 1, wherein each model performing with at least a minimum level of accuracy is applied periodically to the plurality of items, wherein new items and new attributes are associated using new or existing models at various points in time.

3. The computer-implemented method of claim 1, wherein each model is applied only when there is at least a minimum number of data points to be used in determining an association.

4. A computer-implemented method of associating reviewable attributes with items, comprising:

under control of one or more computer systems configured with executable instructions,

identifying a reviewable attribute of a first item, the first item of a plurality of items, the reviewable attribute including a placeholder for receiving text associated with the reviewable attribute of the first item;

determining that the reviewable attribute is associated with at least a minimum number of the plurality of items;

creating a model for the reviewable attribute using the information for the first item associated with the reviewable attribute, the model configured to associate the reviewable attribute with at least a second item such that the second item is configured with a new placeholder;

applying the model to at least a portion of the plurality of items to determine whether the second item should be associated with the reviewable attribute;

when it is determined that the reviewable attribute is to be associated with the second item, associating the reviewable attribute to the second item to enable the new placeholder; and

providing at least the new placeholder based at least in part on the association, the new placeholder configured to enable a user to submit a review for the reviewable attribute associated with the second item.

5. The computer-implemented method of claim 4, wherein the user is enabled to at least view or provide a rating for the reviewable attribute associated with any of the plurality of items.

6. The computer-implemented method of claim 4, further comprising:

submitting one or more mappings of at least one reviewable attribute associated with the at least one item to a structured ratings system for access by the user.

7. The computer-implemented method of claim 4, wherein at least a portion of a group of reviewable attributes were manually associated with at least a portion of the plurality of items.

8. The computer-implemented method of claim 4, wherein at least a portion of a group of reviewable attributes were associated with at least a portion of the plurality of items using a human intelligence task-based system.

9. The computer-implemented method of claim 4, wherein at least a portion of the models is applied iteratively to at least a portion of the plurality of items.

10. The computer-implemented method of claim 4, wherein each model is built using at least one classification algorithm.

11. The computer-implemented method of claim 10, wherein the classification algorithm is one of a voted perceptron algorithm, naï

ve-Bayes classification algorithm, support vector machine (SVM), maximum entropy classification algorithm, decision tree algorithm, or logistic regression algorithm.

12. The computer-implemented method of claim 4, wherein the information includes at least one of review information, item description information, item title information, author information, manufacturer information, origin information, manual information, or information relating to similar items.

13. The computer-implemented method of claim 4, further comprising:

testing the model for each reviewable attribute before applying the model to at least a portion of the plurality of items.

14. The computer-implemented method of claim 13, wherein the model is tested using a k-fold cross-validation process.

15. The computer-implemented method of claim 13, further comprising:

discarding the model when the model does not predict attribute association at or above a threshold level of accuracy.

16. The computer-implemented method of claim 4, wherein creating a model for the reviewable attribute using the information for the first item associated with the reviewable attribute includes building a model using second information that is indicative of an item being associated with an attribute as well as third information that is indicative of an item not being associated with an attribute.

17. The computer-implemented method of claim 4, further comprising:

applying at least one of a time limit or a decay to each association of an item with an attribute, wherein each association is capable of expiring after a period of inactivity.

18. The computer-implemented method of claim 4, further comprising:

pre-processing at least a portion of the information associated at least a portion of the plurality of items using statistical analysis.

19. A system for associating reviewable attributes with items, comprising:

a processor; and

a memory device including instructions that, when executed by the processor, cause the processor to;

identify a reviewable attribute of a first item, the first item of a plurality of items, the reviewable attribute including a placeholder element for receiving review text associated with the reviewable attribute of the first item;

determine that the reviewable attribute is associated with at least a minimum number of the plurality of items;

create a model for the reviewable attribute using the information for the first item associated with the reviewable attribute, the model configured to associate the reviewable attribute with at least a second item such that the second item is configured with a new placeholder element;

apply the model to at least a portion of the plurality of items to determine whether the second item should be associated with the reviewable attribute;

when it is determined that the reviewable attribute is to be associated with the second item, store the association of the reviewable attribute to the second item to enable the new placeholder element; and

in response to a request from a user to review the second item, provide for display at least the new placeholder element configured to enable the user to supply a review for the reviewable attribute associated with the second item.

20. The system according to claim 19, wherein at least a portion of a group of reviewable attributes are manually associated with at least a portion of the plurality of items using at least one of a manual process or a human intelligence task-based system.

21. The system according to claim 19, wherein each model is built using at least one classification algorithm selected from the group consisting of a voted perceptron algorithm, naï

ve-Bayes classification algorithm, support vector machine (SVM), maximum entropy classification algorithm, decision tree algorithm, or logistic regression algorithm.

22. The system according to claim 19, wherein creating a model for each attribute using information for the items associated with the attribute includes building a model using information that is indicative of an item being associated with an attribute as well as information that is indicative of an item not being associated with an attribute.

23. A non-transitory computer readable storage medium including instructions for associating reviewable attributes with items that, when executed by a processor, cause the processor to:

identify a reviewable attribute of a first item, the first item of a plurality of items, the reviewable attribute including a placeholder element for receiving review text associated with the reviewable attribute of the first item;

determine that the reviewable attribute is associated with at least a minimum number of the plurality of items;

create a model for the reviewable attribute using the information for the first item associated with the reviewable attribute, the model configured to associate the reviewable attribute with at least a second item such that the second item is configured with a new placeholder element;

apply the model to at least a portion of the plurality of items to determine whether the second item should be associated with the reviewable attribute;

when it is determined that the reviewable attribute is to be associated with the second item, store the association of the reviewable attribute to the second item to enable the new placeholder element; and

in response to a request from a user to review the second item, provide for display at least the new placeholder element configured to enable the user to supply a review for the reviewable attribute associated with the second item.

24. The non-transitory computer-readable storage medium according to claim 23, wherein each model is built using at least one classification algorithm selected from the group consisting of a voted perceptron algorithm, naï

ve-Bayes classification algorithm, support vector machine (SVM), maximum entropy classification algorithm, decision tree algorithm, or logistic regression algorithm.

25. The non-transitory computer-readable storage medium according to claim 23, wherein creating a model for each attribute using information for the items associated with the attribute includes building a model using information that is indicative of an item being associated with an attribute as well as information that is indicative of an item not being associated with an attribute.