Media recommendations for a social-software website

• US 8,484,226 B2
• Filed: 07/09/2011
• Issued: 07/09/2013
• Est. Priority Date: 07/09/2011
• Status: Active Grant

• Current Assignee: Verizon Patent and Licensing Incorporated (Verizon Communications Inc.)
• Original Assignee: Yahoo! Inc. (Apollo Global Management, Inc.)
• Inventors: Mason, Winter, Suri, Siddharth, Hofman, Jake, Elliott-McCrea, Kellan
• Primary Examiner(s): Le, Miranda
• Application Number:
  US13/179,522
• Publication Number:
  US 20130013622A1
• Time in Patent Office: 731 Days
• Field of Search: 707/751, 707/748, 707/736, 707/999.003
• US Class Current: 707/751
• CPC Class Codes: G06F 16/313 Selection or weighting of terms for indexing
• CPC Class Codes: G06F 16/48 Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
• CPC Class Codes: G06F 16/5866 using information manually generated, e.g. tags, keywords, comments, manually generated location and time information
• CPC Class Codes: G06F 16/587 using geographical or spatial information, e.g. location
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: H04L 51/52 for supporting social networking services

Abstract

Collaborative-filtering software creates a neighborhood using a map-reduce architecture by pair-wise application of a similarity measure to a sparse matrix of users and items of media designated by the users. The collaborative-filtering software then generates recommendations for a particular user by rating items of media designated by other users in the neighborhood (but not the particular user). The collaborative-filtering software rates the item of media, using a weighted vote of the users in the neighborhood. The weighted vote depends at least in part on the mean similarity of the users in the neighborhood who have designated the item of media. Then the collaborative-filtering software records the item of media as a recommendation for subsequent presentation to the other user, if the rating of the item of media is among the highest in comparison to the ratings of other items of media designated by users in the neighborhood.

Claims

1. A method for recommending media on a social-software website, comprising the operations of:

creating a neighborhood using a map-reduce architecture by pair-wise application of a similarity measure to a sparse matrix of users and items of media designated by the users, wherein the sparse matrix is derived from a log, and wherein the measure of similarity is selected from the group consisting of Jaccard similarity, weighted Jaccard similarity, and weighted vote;

identifying an item of media that has been designated by a first user in the neighborhood but not a second user in the neighborhood;

rating the item of media, using a weighted vote of the users in the neighborhood, wherein the weighted vote depends at least in part on the mean similarity of the users in the neighborhood who have designated the item of media; and

recording the item of media as a recommendation for subsequent presentation to the second user in a view in a graphical user interface displayed by a browser, if the rating of the item of media is among the highest in comparison to the ratings of other items of media designated by users in the neighborhood, wherein each operation of the method is executed by one or more processors.

2. The method of claim 1, further comprising the step of receiving as an input a log with a plurality of log entries, wherein each log entry includes an identifier for an item of media, an identifier for a user, and a time when the user designated the item.

3. The method of claim 1, wherein the items of media include photographic images.

4. The method of claim 1, wherein the map-reduce architecture includes a Pig implementation.

5. The method of claim 1, wherein the log or sparse matrix is processed to remove stale entries.

6. The method of claim 1, wherein recording of the second item of media as a recommendation depends on a measure of diversity.

7. The method of claim 1, wherein in a user designates an item of media by bookmarking, marking as a favorite, “

liking,”

or otherwise explicitly indicating an interest in the item of media.

8. The method of claim 1, wherein in a user designates an item of media by viewing it for a particular period of time or otherwise implicitly indicating an interest in the item of media.

9. A computer-readable storage medium persistently storing a program, wherein the program, when executed, instructs a processor to perform the following operations:

create a neighborhood using a map-reduce architecture by pair-wise application of a similarity measure to a sparse matrix of users and items of media designated by the users, wherein the sparse matrix is derived from a log, and wherein the measure of similarity is selected from the group consisting of Jaccard similarity, weighted Jaccard similarity, and weighted vote;

identify an item of media that has been designated by a first user in the neighborhood but not a second user in the neighborhood;

rate the item of media, using a weighted vote of the users in the neighborhood, wherein the weighted vote depends at least in part on the mean similarity of the users in the neighborhood who have designated the item of media; and

record the item of media as a recommendation for subsequent presentation to the second user in a view in a graphical user interface displayed by a browser, if the rating of the item of media is among the highest in comparison to the ratings of other items of media designated by users in the neighborhood.

10. The computer-readable storage medium as in claim 9, further comprising the operation of receiving as an input a log with a plurality of log entries, wherein each log entry includes an identifier for an item of media, an identifier for a user, and a time when the user designated the item.

11. The computer-readable storage medium as in claim 9, wherein the items of media include photographic images.

12. The computer-readable storage medium as in claim 9, wherein the map-reduce architecture includes a Pig implementation.

13. The computer-readable storage medium as in claim 9, wherein the log or sparse matrix is processed to remove stale entries.

14. The computer-readable storage medium as in claim 9, wherein recording of the second item of media as a recommendation depends on a measure of diversity.

15. The computer-readable storage medium as in claim 9, wherein in a user designates an item of media by bookmarking, marking as a favorite, “

liking,”

or otherwise explicitly indicating an interest in the item of media.

16. The computer-readable storage medium as in claim 9, wherein in a user designates an item of media by viewing it for a particular period of time or otherwise implicitly indicating an interest in the item of media.

17. A method for recommending media on a social-software website, comprising the operations of:

maintaining a log with a plurality of log entries, wherein each log entry includes an identifier for an item of media, an identifier for a user, and a time when the user designated the item;

identifying similar users through collaborative filtering of the log, wherein the collaborative filtering employs a software framework based at least in part on a map-reduce architecture and a weighted Jaccard similarity measure;

generating recommendations as to items of media using a weighted vote of the similar users in a neighborhood rating the item of media, wherein the weighted vote depends at least in part on the mean similarity of the similar users in the neighborhood who have designated the item of media; and

presenting the recommendations that have the highest ratings in a view in a graphical user interface displayed by a browser, wherein each operation of the method is executed by one or more processors.