Video-related recommendations using link structure

• US 7,853,622 B1
• Filed: 11/01/2007
• Issued: 12/14/2010
• Est. Priority Date: 11/01/2007
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Sivakumar, Dandapani, Jing, Yushi, Baluja, Shumeet, Yagnik, Jay
• Primary Examiner(s): Wong; Don
• Assistant Examiner(s): LODHI, ANDALIB FT
• Application Number:
  US11/933,676
• Time in Patent Office: 1,139 Days
• Field of Search: None
• US Class Current: 707/803
• CPC Class Codes: G06F 16/437 Administration of user profiles, e.g. generation, initialisation, adaptation, distribution
• CPC Class Codes: G06F 16/735 Filtering based on additional data, e.g. user or group profiles
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: Y10S 707/99942 Manipulating data structure, e.g. compression, compaction, compilation
• CPC Class Codes: Y10S 707/99945 Object-oriented database structure processing

Abstract

The subject matter of this specification can be embodied in, among other things, a method that includes inferring labels for videos, users, advertisements, groups of users, and other entities included in a social network system. The inferred labels can be used to generate recommendations such as videos or advertisements in which a user may be interested. Inferred labels can be generated based on social or other relationships derived from, for example, profiles or activities of social network users. Inferred labels can be advantageous when explicit information about these entities is not available. For example, a particular user may not have clicked on any online advertisements, so the user is not explicitly linked to any advertisements.

Claims

1. A computer-implemented method comprising:

receiving, at a computer system, user information associated with users of a social network and associated with media accessed by the users;

generating, by the computer system, a graph having nodes of a plurality of categories that are linked together by edges from the received user information, wherein the plurality of categories includes at least a first category of user nodes that represent the users of the social network and a second category of media nodes that represent the media accessed by the users;

selecting, by the computer system, a type of label to propagate in the graph;

iteratively propagating, by the computer system, values for labels of the selected label type through nodes of the plurality of categories in the graph using the edges that link the nodes together; and

outputting by the computer system, the iteratively propagated values for labels of the selected label type for the nodes of the plurality of categories in the graph.

2. The method of claim 1, wherein the user information associated with the users of the social network comprises user identifiers, group identifiers for groups of users, demographic information for the users, and activity indicators that specify how active users are on the social network.

3. The method of claim 1, wherein the user information associated with the media accessed by the users includes information about videos watched by users, genres associated with the videos, advertisements (ads) selected by or displayed to users, email communications sent or received by the users, and comments posted or read by the users.

4. The method of claim 1, wherein the selected label type comprises videos watched by users.

5. The method of claim 4, wherein outputting a value of a label of the selected label type for a particular user node comprises determining a value for a particular video label associated with a particular user represented by the particular user node.

6. The method of claim 5, further comprising determining one or more videos to recommend to the particular user by identifying one or more videos labels having values for the particular user node that satisfy a threshold requirement.

7. The method of claim 1, wherein the selected label type comprises ads selected by or displayed to the users.

8. The method of claim 7, wherein outputting a value of a label of the selected label type for a particular user node comprises determining a value for a particular ad label associated with a particular user represented by the particular user node.

9. The method of claim 8, further comprising determining one or more ads to recommend for display to the particular user by identifying one or more ad labels having values for the particular user node that satisfy a threshold requirement.

10. The method of claim 7, wherein outputting a value of a label of the selected label type for a particular media node comprises determining a value for a particular ad label associated with a particular video represented by the particular media node.

11. The method of claim 10, further comprising determining one or more ads to recommend for display with the particular video by identifying one or more ad labels having values for the particular media node that satisfy a threshold requirement.

12. The method of claim 1, wherein edges that link user nodes together in the graph are derived from relationships among users represented by the user nodes, wherein the relationships comprise friendships specified on user profiles, common membership in communities or groups on the social network, common genre classifications, common ads selected by the users, common classifications inferred from demographic information specified on user profiles, inferred relationships based on a first user posting content to a second user'"'"'s web page, or relationships inferred from users'"'"' access to media.

13. The method of claim 1, wherein edges that link user nodes and media nodes together in the graph are weighted based on user ratings of the media or a number of times users accessed the media, a correlation of media to a genre based on user or online database ratings, whether video media includes advertising media, a correlation of video media to advertising media based on advertisers'"'"' input or the video media authors'"'"' input, whether users list each other as friends on a social network, or how active a particular user is on the social network.

14. The method of claim 13, wherein the weighting affects a determination of the values for labels of the selected label type.

15. The method of claim 1, wherein iteratively propagating the values for the labels of the selected label type through the nodes comprises performing a random walk algorithm on the graph to determine the values for the labels of the selected label type.

16. The method of claim 1, wherein iteratively propagating the values for the labels of the selected label type through the nodes comprises performing an adsorption algorithm on the graph to determine the values for the labels of the selected label type.

17. The method of claim 1, further comprising normalizing, for a particular node, values for labels of the selected label type iteratively propagated to the particular node.

18. The method of claim 1, wherein the type of label is selected from a plurality of types of labels that are each associated with one or more categories from the plurality of categories.

19. A computer-implemented method comprising:

receiving user information associated with users of a social network and associated with videos accessed by the users;

generating a graph that links graph nodes based on relationships derived from the user information, the graph nodes comprising

user nodes representing the users,

ad nodes representing advertisements, and

video nodes representing the videos;

selecting one or more of the user, ad, or video nodes as a label to associate with the graph nodes;

wherein the user nodes make up a user node group, the ad nodes make up an ad node group, and the video nodes make up a video node group, and at least some of the nodes in one of the groups are linked by edges in the graph to other of the nodes in another of the groups; and

outputting, for a first graph node, a first label value for the label associated with the particular graph node based on one or more second label values for the label that are associated with neighboring graph nodes that are linked to the particular graph node.

20. A system comprising:

one or more servers;

an interface to the one or more servers that is configured to receive social network information associated with users of a social network and media accessed by the users;

one or more processors of the one or more servers that are configured to execute instructions; and

a computer-readable storage medium accessible to the one or more servers that is configured to store instructions, wherein execution of the stored instructions by the one or more processors causes the one or more servers to;

generate a graph having nodes of a plurality of categories that are linked together by edges from the social network information, wherein the plurality of categories includes at least a first category of user nodes that represent the users of the social network and a second category of media nodes that represent the media accessed by the users;

select a type of label to propagate in the graph;

iteratively propagate values for labels of the selected label type through nodes of the plurality of categories in the graph using the edges that link the nodes together; and

output the iteratively propagated values for labels of the selected label type for the nodes of the plurality of categories in the graph.

21. A computer-implemented method comprising:

identifying, by a computer system, a relationship between a first user and a second user of a social network using a graph having nodes of a plurality of categories that are linked together by edges, wherein the plurality of categories includes at least a first category of user nodes that represent the users of the social network and a second category of media nodes that represent the media accessed by the users, and wherein a first node representing the first user is linked to a second node representing the second user by an edge in the graph;

associating, by the computer system, first information used to target advertising with the first node based on actions or content generated by the first user at one or more systems external to the social network, wherein at least some of the first information is associated with the first node by an edge linking a media node to the first node, wherein the first information comprises one or more labels that are selected for propagating in the graph and that have values that are propagated in the graph through the nodes of the graph using the edges that link the nodes together;

inferring, by the computer system based on the first information and the relationship, second information used to target advertising to the second user of the social network; and

outputting, by the computer system, an advertisement for the second user based on the inferred second information.

22. The method of claim 21, wherein the relationship includes a friendship relationship.