Method, system, and apparatus for determining a predicted rating

• US 8,732,104 B2
• Filed: 10/05/2009
• Issued: 05/20/2014
• Est. Priority Date: 10/03/2008
• Status: Expired due to Fees

• Current Assignee: Sift LLC
• Original Assignee: Sift LLC
• Inventors: Pohl, Daniel
• Primary Examiner(s): Chaki, Kakali
• Assistant Examiner(s): SITIRICHE, LUIS A
• Application Number:
  US12/573,648
• Publication Number:
  US 20100088265A1
• Time in Patent Office: 1,688 Days
• Field of Search: None
• US Class Current: 706/46
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising

Abstract

Embodiments of the invention can incorporate a method and interface to accomplish one or more of the following three processes: input, analysis, and output. With respect to input, the user(s) identifier(s), item(s) identifier(s), and user(s)-item(s) rating data 22 can be accepted through an interface 23 from an application 21. With respect to analysis, the method and interface can then take the actual user-item ratings 31 and process 32 the data to generate and store item predictor information 33. The predicted user-item ratings 41 and the actual user-item ratings 42 can then be outputted by, for example, passing these outputs to the application via an interface 44.

Claims

1. A method of determining a predicted rating, comprising:

receiving a plurality of data sets, wherein each data set comprises a user identifier, an item identifier, and a rating, wherein for each data set each user identifier identifies a user of a plurality of users, each item identifier identifies an item of a plurality of items, and each rating is a rating by the user of the plurality of users identified by the user identifier in the data set and of the item of the plurality of items identified by the item identifier in the data set;

identifying, via a suitably programmed computer, an identified pair of items of the plurality of items for which a rating for one item in the identified pair of items for any user of the plurality of users can be used to predict a predicted rating for the other item in the identified pair of items for the same user of the plurality of users, based on an analysis of ratings including ratings for the one item in the identified pair of items and ratings for the other item in the identified pair of items, wherein the ratings for the one item in the identified pair of items and the ratings for the other item in the identified pair of items include ratings from users of the plurality of users having ratings for only one item in the identified pair of items; and

determining the predicted rating for a given item of the identified pair of items for a given user of the plurality of users based on a given rating of the other item of the identified pair of items for the given user of the plurality of users.

2. The method according to claim 1, wherein identifying the identified pair of items comprises identifying an identified pair of items having ratings that are highly correlated.

3. The method according to claim 2, wherein identifying the identified pair of items having ratings that are highly correlated comprises calculating a statistical variance of a set of ratings, wherein the set of ratings includes ratings of the_items in the identified pair of items.

4. The method according to claim 3, wherein the set of ratings includes all ratings in the plurality of data sets of the items in the identified pair of items.

5. The method according to claim 3, wherein the set of ratings includes ratings from users of the plurality of users having ratings for both items in the identified pair of items.

6. The method according to claim 3, wherein identifying the identified pair of items having ratings that are highly correlated comprises finding that the statistical variance of the set of ratings is less than a threshold value.

7. The method according to claim 3, wherein identifying the identified pair of items having ratings that are highly correlated comprises finding that the statistical variance of the set of ratings is less than a statistical variance of a second set of ratings including of ratings of items from other pairs of items.

8. The method according to claim 3, wherein calculating the statistical variance of the set of ratings comprises using a statistical method selected from the group consisting of standard deviation, Horting, Pearson, Pearson product-moment correlation coefficient, Pearson R, Pearson T, Spearman'"'"'s rank correlation coefficient, Kendall'"'"'s tau rank correlation coefficient, Confounding variable, and Point-Biserial Correlation.

9. The method according to claim 1, wherein the analysis of ratings including ratings for the one item in the identified pair or items and ratings for the other item in the identified pair of items is an analysis of ratings including all ratings in the plurality of data sets of the items in the identified pair of items.

10. The method according to claim 1, further comprising:

(a) identifying, via a suitably programmed computer, an additional identified pair of items for which a rating for one item in the additional identified pair of items for any user of the plurality of users can be used to predict an additional predicted rating for the other item in the additional identified pair of items for the same user of the plurality of users, based on an analysis of ratings including ratings for the one item in the additional identified pair of items and ratings for the other item in the additional identified pair of items, wherein the ratings for the one item in the additional identified pair of items and the ratings for the other item in the additional identified pair of items include ratings from users of the plurality of users having ratings for only one item in the additional identified pair of items;

(b) determining the additional predicted rating for a given item of the additional identified pair of items for a given user of the plurality of users based on a given rating of the other item of the additional identified pair of items for the given user of the plurality of users; and

(c) repeating (a) and (b) until all predicted ratings for all combinations of users of the plurality of users and items of the plurality of items has been determined.

11. The method according to claim 10, further comprising:

storing all predicted ratings.

12. The method according to claim 10, further comprising:

outputting all predicted ratings.

13. The media according to claim 12, wherein the method further comprises:

storing all predicted ratings.

14. The method according to claim 1, further comprising:

using the rating for the one item in the identified pair of items for a first user of the plurality of users to predict the predicted rating for the other item in the identified pair of items for the first user of the plurality of users.

15. One or more non-transitory computer-readable media having computer-useable instructions embodied thereon for performing a method of determining a predicted rating, the method comprising:

receiving a plurality of data sets, wherein each data set comprises a user identifier, an item identifier, and a rating, wherein for each data set each user identifier identifies a user of a plurality of users, each item identifier identifies an item of a plurality of items, and each rating is a rating by the user of the plurality of users identified by the user identifier in the data set of the item identified by the item identifier in the data set;

identifying an identified pair of items for which a rating for one item in the identified pair of items for any user of the plurality of users can be used to predict a predicted rating for the other item in the identified pair of items for the same user of the plurality of users, based on an analysis of ratings including ratings for the one item in the identified pair of items and ratings for the other item in the identified pair of items, wherein the ratings for the one item in the identified pair of items and the ratings for the other item in the identified pair of items include ratings from users of the plurality of users having ratings for only one item in the identified pair of items; and

determining the predicted rating for a given item of the identified pair of items for a given user of the plurality of users based on a given rating of the other item of the identified pair of items for the given user of the plurality of users.

16. The media according to claim 15, wherein identifying the identified pair of items comprises identifying an identified pair of items having ratings that are highly correlated.

17. The media according to claim 16, wherein identifying the identified pair of items having ratings that are highly correlated comprises calculating a statistical variance of a set of ratings, wherein the set of ratings includes ratings of the items in the identified pair of items.

18. The media according to claim 17, wherein the set of ratings includes all ratings in the plurality of data sets of the items in the identified pair of items.

19. The media according to claim 17, wherein the set of ratings includes ratings from users of the plurality of users having ratings for both items in the identified pair of items.

20. The media according to claim 17, wherein identifying the identified pair of items having ratings that are highly correlated comprises finding that the statistical variance of the set of ratings is less than a threshold value.

21. The media according to claim 17, wherein identifying the identified pair of items having ratings that are highly correlated comprises finding that the statistical variance of the set of ratings is less than a statistical variance of a second set of ratings including of ratings of items from other pairs of items.

22. The media according to claim 17, wherein calculating the statistical variance of the set of ratings comprises using a statistical method selected from the group consisting of standard deviation, Horting, Pearson, Pearson product-moment correlation coefficient, Pearson R, Pearson T, Spearman'"'"'s rank correlation coefficient, Kendall'"'"'s tau rank correlation coefficient, Confounding variable, and Point-Biserial Correlation.

23. The media according to claim 15, wherein the analysis of ratings including ratings for the one item in the identified pair of items and ratings for the other item in the identified pair of items is an analysis of ratings including all ratings in the plurality of data sets of the items in the identified pair of items.

24. The media according to claim 15, wherein the method further comprises:

(a) identifying, via a suitably programmed computer, an additional identified pair of items for which a rating for one item in the additional identified pair of items for any user of the plurality of users can be used to predict an additional predicted rating for the other item in the additional identified pair of items for the same user of the plurality of users, based on an analysis of ratings including ratings for the one item in the additional identified pair of items and ratings for the other item in the additional identified pair of items, wherein the ratings for the one item in the additional identified pair of items and the ratings for the other item in the additional identified pair of items include ratings from users of the plurality of users having ratings for only one item in the additional identified pair of items;

(b) determining the additional predicted rating for a given item of the additional identified pair of items for a given user of the plurality of users based on a given rating of the other item of the additional identified pair of items for the given user of the plurality of users; and

(c) repeating (a) and (b) until all predicted ratings for all combinations of users of the plurality of users and items of the plurality of items has been determined.

25. The media according to claim 24, wherein the method further comprises:

outputting all predicted ratings.

26. A system for providing a predicted rating, comprising:

an input interface, wherein the input interface is adapted to receive a plurality of data sets, wherein each data set comprises a user identifier, an item identifier, and a rating, wherein for each data set each user identifier identifies a user of a plurality of users, each item identifier identifies one of a plurality of items, and each rating is a rating by the user of the plurality of users identified by the user identifier in the data set of the item identified by the item identifier in the data set;

a predictive pair identification module, wherein the predictive pair identification module identifies an identified pair of items for which a rating for one item in the identified pair of items for any user of the plurality of users can be used to predict a rating for the other item in the identified pair of items for the same user of the plurality of users, based on an analysis of ratings including ratings for the one item in the identified pair of items and ratings for the other item in the identified pair of items, wherein the ratings for the one item in the identified pair of items and the ratings for the other item in the identified pair of items include ratings from users of the plurality of users having ratings for only one item in the identified pair of items; and

a predictive rating determination module, wherein the predictive rating determination module determines a predicted rating for a given item of the identified pair of items for a given user of the plurality of users based on a given rating of the other item of the identified pair of items for the given user of the plurality of users;

one or more computer readable media having computer-useable instructions embodied thereon for performing the functions of the predictive pair identification module and the predictive rating determination module;

one or more processors for executing the computer-useable instructions on the one or more computer readable media; and

an output interface, wherein the output interface presents the predicted rating.

27. The system according to claim 26, further comprising:

a memory for storing one or more of the following;

some or all of the plurality of data sets;

the identified pair of items; and

the predicted rating.

28. The system according to claim 27, further comprising an output device, wherein the output interface presents the predicted rating via the output device.

29. The system according to claim 28, wherein the output device comprises a computer monitor.

30. The system according to claim 28, wherein the output device comprises a communication device that communicates the predicted rating to another computer system.

31. The system according to claim 30, wherein the output device comprises a modem that communicates the predicted rating to a computer via a network.

32. The system according to claim 26, wherein the input interface comprises:

the plurality of data sets to a computer program for performing the functions of the predictive pair indentification module and the predictive rating determination module, the interface consisting essentially of;

a first input field, wherein the first input field receives the user identifier of each data set;

a second input field, wherein the second input field receives the item identifier of each data set; and

a third input field, wherein the third input field receives the item rating of each data set.

33. The system according to claim 26,

wherein the predictive pair identification module (a) identifies an additional pair of items for which a rating for one item in the additional identified pair of items for any user of the plurality of users can be used to predict an additional predicted rating for the other item in the additional identified pair of items for the same user of the plurality of users, based on an analysis of ratings including ratings for the one item in the additional identified pair of items and ratings for the other item in the additional identified pair of items, wherein the ratings for the one item in the additional identified pair of items and the ratings for the other item in the additional identified pair of items include ratings from users of the plurality of users having ratings for only one item in the additional identified pair of items,

wherein the predictive rating determination module (b) determines the additional predicted rating for a given item of the additional identified pair of items for a given user of the plurality of users based on a given rating of the other item of the additional identified pair of items for the given user of the plurality of users,

wherein the predictive pair identification module repeats (a) and the predictive rating determination module repeats (b) until all predicted ratings for all combinations of users of the plurality of users and items of the plurality of items has been determined.

34. The system according to claim 33, further comprising:

a storage module, wherein the storage module stores all predicted ratings.

35. The system according to claim 33, wherein the output interface presents all predicted ratings.