MEDIA CONTENT RANKINGS FOR DISCOVERY OF NOVEL CONTENT

• US 20150039608A1
• Filed: 07/30/2013
• Published: 02/05/2015
• Est. Priority Date: 07/30/2013
• Status: Active Grant

• Current Assignee: Netflix, Inc.
• Original Assignee: Netflix.Com, Inc.
• Inventors: Basilico, Justin D.
• Granted Patent: US 9,430,532 B2
• Time in Patent Office: Days
• US Class Current: 707/734
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/735 Filtering based on additional data, e.g. user or group profiles
• CPC Class Codes: G06N 20/00 Machine learning

Abstract

A content provider system ranks media content items with respect to a particular user based on selection scores determined for each of the media content items. The selection scores may be determined using a particular model that calculates a predicted selection score based on feature values associated with the content item with respect to the particular user. The feature values may indicate properties of the media content item, the particular user, or the particular user'"'"'s relationship with the content item, including information about the novelty of the media content item with respect to the user. The particular model may be trained with sample user consumption data points that represent various combinations of media content items and users. The data point information evaluated during the training of the particular model may cause the model to assign higher selection scores to content items that are novel in particular ways.

Claims

1. A method comprising:

in a data processing system configured to generate personalized rankings of input media content items in a catalog;

determining a plurality of user consumption values, wherein each of the user consumption values is associated with a user consumption data point, and wherein each user consumption data point is associated with a corresponding user and a corresponding media content item;

wherein each of the user consumption values indicate a user consumption value with respect to a user that corresponds to the user consumption data point associated with the user consumption value; and

a media content item that corresponds to the data point associated with the user consumption value;

determining feature values for each of the user consumption data points associated with the plurality of user consumption values, wherein the feature values include a particular feature value for a particular user consumption data point indicating novelty information for a particular media content item associated with the particular user consumption data point with respect to a particular user associated with the particular user consumption data point;

determining, based on the plurality of user consumption values and the feature values, predicted selection scores for an input user and the input media content items;

ranking the media content items in the catalog for the input user based on the predicted selection scores;

wherein the method is performed using one or more processors.

2. The method of claim 1, wherein a certain user consumption value associated with a certain user consumption data point is determined based upon:

certain novelty information indicating novelty of a certain content item corresponding to the certain user consumption data point with respect to a certain user corresponding to the particular user consumption data point;

user consumption information indicating whether the certain content item was selected for consumption by the certain user.

3. The method of claim 2, wherein the novelty of the certain content item with respect to the certain user is determined based on novelty of a related content item with respect to the certain user, wherein the related content item is different than the certain content item.

4. The method of claim 2, wherein the certain novelty information indicates novelty of a similar content item with respect to the certain user, wherein the similar content item is different than the certain content item and is determined to be similar to the certain content item.

5. The method of claim 2, wherein the certain novelty information indicates at least one of:

whether the certain content item was consumed for at least a particular amount of time;

whether the certain content item was reviewed;

whether an option to consume the content item was displayed to the input user;

whether the certain content item was rated;

or

whether the certain content item was added to queue of content items for consumption.

6. The method of claim 2, wherein the certain novelty information is a non-binary novelty value indicating a level of novelty of the certain content item with respect to the certain user.

7. The method of claim 2, wherein the certain user consumption value is a certain binary value indicating that the certain content item is novel with respect to the certain user and that the certain user selected the certain content item for consumption.

8. The method of claim 1, wherein the feature values further include an additional feature value for the particular user consumption data point indicating further novelty information about the particular media content item with respect to the particular user, wherein the further novelty information is different than the novelty information indicated by the particular feature value.

9. The method of claim 1, wherein the feature values further include an additional feature value for the particular user consumption data point indicating popularity of the particular media content item with respect to users other than the particular user.

10. The method of claim 1, wherein determining the predicted selection scores comprises determining a regression function based on the feature values and the plurality of user consumption values, and wherein the regression function is an expression for calculating a predicted selection score for a corresponding input media content item and a corresponding input user.

11. The method of claim 1, wherein determining the predicted selection scores comprises determining a classification function based on the feature values and the plurality of user consumption values, and wherein the classification function is an expression for calculating a predicted selection score for a corresponding input media content item and a corresponding input user.

12. The method of claim 2, wherein determining the predicted selection scores comprises using a computational expression for calculating an output predicted selection score based on a set of input feature values.

13. The method of claim 12, wherein the computational expression identifies a plurality of weight values, wherein each weight value of the plurality of weight values corresponds to a respective feature and indicates a relative weight to be applied to an input feature value of the respective feature during calculation of the output predicted selection score.

14. The method of claim 1, further comprising causing display of media content information representing the media content items, wherein the media content information is ordered according to a result of the ranking.

15. The method of claim 1, wherein the user consumption data points, with which the plurality of user consumption values is associated, represent observed user consumption of media content items.

16. The method of claim 1, further comprising:

determining an unfavorable user consumption value for a certain user consumption data point associated with a certain user and a certain content item in response to determining;

that the certain content item was selected for consumption by the certain user; and

that the certain content item was not novel with respect to the certain user.

17. A data processing system for generating personalized ranking of media content items in a catalog, the system comprising one or more computers configured to:

determine a plurality of user consumption values, wherein each of the user consumption values is associated with a user consumption data point, and wherein each user consumption data point is associated with a corresponding user and a corresponding media content item;

wherein each of the user consumption values indicate a user consumption value with respect to a user that corresponds to the user consumption data point associated with the user consumption value; and

a media content item that corresponds to the data point associated with the user consumption value;

determine feature values for each of the user consumption data points associated with the plurality of user consumption values, wherein the feature values include a particular feature value for a particular user consumption data point indicating novelty information for a particular media content item associated with the particular user consumption data point with respect to a particular user associated with the particular user consumption data point;

determine, based on the plurality of user consumption values and the feature values, predicted selection scores for an input user and an input media content item;

rank the media content items in the catalog for the input user based on the predicted selection scores.

18. The data processing system of claim 17, wherein a certain user consumption value associated with a certain user consumption data point is determined based upon:

certain novelty information indicating novelty of a certain content item corresponding to the certain user consumption data point with respect to a certain user corresponding to the particular user consumption data point;

user consumption information indicating whether the certain content item was selected for consumption by the certain user.

19. The data processing system of claim 18, wherein the novelty of the certain content item with respect to the certain user is determined based on novelty of a related content item with respect to the certain user, wherein the related content item is different than the certain content item.

20. The data processing system of claim 18, wherein the certain novelty information indicates novelty of a similar content item with respect to the certain user, wherein the similar content item is different than the certain content item and is determined to be similar to the certain content item.

21. The data processing system of claim 18, wherein the certain novelty information indicates at least one of:

whether the certain content item was consumed for at least a particular amount of time;

whether the certain content item was reviewed;

whether an option to consume the content item was displayed to the input user;

whether the certain content item was rated;

or

whether the certain content item was added to queue of content items for consumption.

22. The data processing system of claim 18, wherein the certain novelty information is a non-binary novelty value indicating a level of novelty of the certain content item with respect to the certain user.

23. The data processing system of claim 18, wherein the certain user consumption value is a certain binary value indicating that the certain content item is novel with respect to the certain user and that the certain user selected the certain content item for consumption.

24. The data processing system of claim 17, wherein the feature values further include an additional feature value for the particular user consumption data point indicating further novelty information about the particular media content item with respect to the particular user, wherein the further novelty information is different than the novelty information indicated by the particular feature value.

25. The data processing system of claim 17, wherein the feature values further include an additional feature value for the particular user consumption data point indicating popularity of the particular media content item with respect to users other than the particular user.

26. The data processing system of claim 17, wherein determining the predicted selection scores comprises determining a regression function based on the feature values and the plurality of user consumption values, and wherein the regression function is an expression for calculating a predicted selection score for a corresponding input media content item and a corresponding input user.

27. The data processing system of claim 17, wherein determining the predicted selection scores comprises determining a classification function based on the feature values and the plurality of user consumption values, and wherein the classification function is an expression for calculating a predicted selection score for a corresponding input media content item and a corresponding input user.

28. The data processing system of claim 17, wherein determining the predicted selection scores comprises using a computational expression for calculating an output predicted selection score based on a set of input feature values.

29. The data processing system of claim 28, wherein the computational expression identifies a plurality of weight values, wherein each weight value of the plurality of weight values corresponds to a respective feature and indicates a relative weight to be applied to an input feature value of the respective feature during calculation of the output predicted selection score.

30. The data processing system of claim 17, further configured to cause display of media content information representing the media content items, wherein the media content information is ordered according to a result of the ranking.

31. The method of claim 17, wherein the user consumption data points, with which the plurality of user consumption values is associated, represent observed user consumption of media content items.

32. The data processing system of claim 17, further configured to:

determine an unfavorable user consumption value for a certain user consumption data point associated with a certain user and a certain content item in response to determining;

that the certain content item was selected for consumption by the certain user; and

that the certain content item was not novel with respect to the certain user.