GENERATING RECOMMENDATIONS THROUGH USE OF A TRUSTED NETWORK

• US 20100312644A1
• Filed: 06/04/2009
• Published: 12/09/2010
• Est. Priority Date: 06/04/2009
• Status: Abandoned Application

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Borgs, Christian Herwarth, Chayes, Jennifer Tour, Kalai, Adam Tauman, Body, Danah Michele, Tennenholtz, Moshe
• Application Number:
  US12/477,914
• Publication Number:
  US 20100312644A1
• Time in Patent Office: Days
• US Class Current: 705/14.55
• CPC Class Codes: G06N 20/10 using kernel methods, e.g. support vector machines [SVM]
• CPC Class Codes: G06N 3/02 Neural networks
• CPC Class Codes: G06N 5/022 Knowledge engineering; Knowledge acquisition
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks
• CPC Class Codes: G06Q 10/10 Office automation; Time management
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0257 User requested
• CPC Class Codes: G06Q 30/0282 Rating or review of business operators or products
• CPC Class Codes: G06Q 50/01 Social networking

Abstract

Described herein are various techniques for automatically generating recommendations for a user based upon a social network of the user. A user can request a recommendation for a particular context, and a social network of the user can be automatically pared to create a subnetwork, wherein individuals in the subnetwork have provided ratings for the particular context and/or contexts that are in some way related to the particular context. The knowledge/ratings of individuals in the subnetwork may then be leveraged to automatically generate the recommendation for the particular context.

Claims

1. A method comprising the following computer-executable acts:

receiving a request to generate a recommendation for a user for a particular context;

a) accessing a data repository in a computing device that comprises;

i) a first data set that is representative of a first graph that includes a plurality of nodes and a plurality of edges that connect such nodes, wherein the graph is representative of a social network of the user, wherein the nodes in the graph are representative of individuals in the social network of the user and edges in the graph represent trusted relationships between individuals represented by nodes;

ii) ratings provided by one or more of the individuals in the social network with respect to a plurality of contexts;

b) automatically removing at least one node and at least one edge from the graph to create a second data set based at least in part upon the particular context and the ratings, wherein the second data set is representative of a second graph, wherein the second graph is representative of a subnetwork of the social network; and

c) automatically generating the recommendation for the user for the particular context based at least in part upon the subnetwork of the social network and rankings for various contexts generated by individuals in the subnetwork of the social network.

2. The method of claim 1, wherein nodes in the first graph are representative of individuals in at least one of:

an email contact list of the user;

a social networking website contact list of the user;

a telephone contact list of the user;

or

an instant messaging contact list of the user.

3. The method of claim 1, wherein the particular context is a pre-defined context selected from a graphical user interface by the user.

4. The method of claim 1, wherein edges of the first graph are weighted to indicate a level of trust between the user and individuals in the social network of the user.

5. The method of claim 4, further comprising:

receiving feedback from the user with respect to the recommendation; and

automatically modifying a weight of at least one edge in the first graph based at least in part upon the feedback of the user.

6. The method of claim 1, wherein automatically removing the at least one node and the at least one edge from the graph to create the second data set comprises utilizing a supervised learning technique to determine which nodes and which edges to remove from the first graph.

7. The method of claim 1, wherein automatically removing the at least one node and the at least one edge from the graph to create the second data set comprises:

receiving a learned relationship between contexts; and

automatically removing the at least one node and the at least one edge from the graph based at least in part upon the learned relationship between contexts.

8. The method of claim 1, wherein automatically removing the at least one node and the at least one edge from the graph to create the second data set comprises using a leave one out estimation approach in connection with determining which nodes and which edges to remove from the first graph.

9. The method of claim 8, wherein automatically removing the at least one node and the at least one edge from the graph to create the second data set comprises:

assigning a penalty value for removing an edge from the first graph;

determining an error value when removing the edge from the first graph; and

removing the edge from the first graph based at least in part upon a combination of the error value and the penalty value with respect to the edge.

10. The method of claim 1, further comprising monetizing the recommendation.

11. The method of claim 10, wherein the output recommendation is in the form of an advertisement.

12. The method of claim 1, wherein the social network comprises contacts that are at least one degree of separation from the user.

13. The method of claim 1, wherein the first graph is representative of an aggregation of data from multiple social network sources pertaining to the user.

14. A system comprising the following computer-executable components:

a network parer component that receives;

a first data set that is representative of a first graph, wherein the first graph is representative of a social network of a user, wherein the first graph comprises nodes that are representative of individuals in the social network of the user and edges that represent trusted relationships between individuals in the first graph;

ratings from a plurality of individuals in the social network of the user pertaining to various contexts; and

a request to generate a recommendation for the user for a particular context, wherein the network parer component automatically pares the first graph to remove at least one edge from the first graph to create a second graph that is representative of a subnetwork of the social network, wherein the network parer component automatically pares the first graph based at least in part upon the ratings from the plurality of individuals in the social network of the user and the particular context; and

a recommender component that generates the recommendation based at least in part upon the second graph.

15. The system of claim 14, wherein the network parer component receives learned relationships between contexts and automatically pares the first graph to remove at least one edge from the first graph based at least in part upon the learned relationships between contexts.

16. The system of claim 14, wherein the particular context is at least one of time-dependent or location-dependent.

17. The system of claim 14, further comprising an updater component that receives feedback from the user pertaining to the recommendation and automatically updates the subnetwork based at least in part upon the received feedback.

18. The system of claim 14, wherein edges of the first graph are weighted, and wherein weights of edges in the first graph are modified depending on a context of interest.

19. The system of claim 14, wherein weights of edges in the first graph are modified based upon feedback from the user pertaining to the generated recommendation.

20. A computer-readable medium comprising instructions that, when executed by a processor, cause the processor to perform the following acts:

receiving a first data set that is representative of a first weighted graph, wherein the first weighted graph comprises nodes and weighted, directional edges that indicate trusted relationships between the nodes, wherein the first weighted graph is representative of a social network of the user, wherein the nodes of the first weighted graph represent individuals in the social network of the user and the weighted, directional edges represent trusted relationships between individuals in the social network, and wherein the first data set comprises data from multiple social network sources;

receiving a request for a recommendation for a particular context;

receiving a plurality of ratings for a plurality of different contexts generated by individuals in the social network of the user;

automatically modifying the first data set to generate a second data set that is representative of a second weighted graph, wherein the second weighted graph is representative of subnetwork of the social network, wherein the first data set is modified by removing at least one edge of the first weighted graph to generate the second weighted graph, wherein the first data set is automatically modified based at least in part upon the received plurality of ratings and the particular context; and

executing a recommendation algorithm over the second data set to generate the recommendation for the particular context.