Trust-based recommendation systems

• US 7,991,841 B2
• Filed: 04/28/2008
• Issued: 08/02/2011
• Est. Priority Date: 10/24/2007
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Mirrokni, Seyed Vahab, Borgs, Christian Herwarth, Flaxman, Abraham, Chayes, Jennifer Tour, Feige, Uriel Mordechai, Kalai, Adam Tauman, Anderson, Reid M., Tennenholtz, Moshe
• Primary Examiner(s): Chan; Wing F
• Assistant Examiner(s): Yohannes; Tesfay
• Application Number:
  US12/110,625
• Publication Number:
  US 20090112989A1
• Time in Patent Office: 1,191 Days
• Field of Search: 709/205, 709/206, 709/213
• US Class Current: 709/205
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising

Abstract

Systems and methods that analyze aggregated item evaluation behavior of users, to suggest a recommendation for the item. An analysis component forms a collective opinion by taking as input votes of users and trusted relationships established therebetween, to output an evaluation and/or recommendation for the item. Accordingly, within a linked structure of nodes, personalized recommendations to users (e.g., agents) are supplied about an item(s) based upon the opinions/reviews of other users, and in conjunction with the declared trust between the users.

Claims

1. A computer implemented system comprising:

at least one processor;

a network of nodes and edges, each of the nodes representing a user of a plurality of users in a social network and each of the edges representing a trust relationship between users of the plurality of users;

the following computer executable components, executable by the at least one processor;

a component that receives trust relationships declared between users of the plurality of users, thereby maintaining the network of nodes and edges; and

an analysis component associated with the network to supply an aggregate recommendation for an item to a first user of the plurality of users, the aggregate recommendation being based in part on votes relating to the item designated by a portion of the users, the portion comprising one or more second users selected such that the network comprises a path along the edges of the network from a node representing the first user to a node representing the second user, the analysis component aggregating the votes based in part on trust relationships among the first user and the one or more second users.

2. The computer implemented system of claim 1, the votes including a “

yes”

vote, or a “

no”

vote, or a “

neutral”

vote.

3. The computer implemented system of claim 2 further comprising a machine learning system that infers relationships among the votes and the trust relationships.

4. The computer implemented system of claim 2 further comprising an axiomatic property that defines class models.

5. The computer implemented system of claim 4 wherein the network comprises a directed acyclic graph.

6. The computer implemented system of claim 4, wherein the analysis component comprises a Random Walk System (FRW) system.

7. The computer implemented system of claim 4, wherein the analysis component comprises a Majority-of-Majorities (FMoM) system.

8. The computer implemented system of claim 4, wherein the analysis component applies a positive response axiom.

9. The computer implemented system of claim 4, wherein the analysis component comprises a Min-cut system (Fmincut).

10. A computer implemented method of supplying a recommendation for an item comprising:

with at least one processor;

receiving user input declaring trust relationships between a plurality of nodes in a network;

receiving user input supplying a vote(s) by a node regarding an item;

receiving user input identifying a node of the plurality of nodes; and

designating a collective opinion of the plurality of nodes for the selected node, the collective opinion being based on the trust relationships and the vote(s), the designating comprising combining, based on the declared trust relationships, votes regarding the item by nodes linked to the selected node through the declared trust relationships.

11. The computer implemented method of claim 10 further comprising defining class models for the network.

12. The computer implemented method of claim 10 further comprising identifying a set of natural axioms for the network.

13. The computer implemented method of claim 10, further comprising supplying a Directed Acylic Graph and partitioning the node into a finite number of levels.

14. The computer implemented method of claim 12 further comprising designating the set of natural axioms.

15. The computer implemented method of claim 12 further comprising outputting a recommendation for the item.

16. The computer implemented method of claim 12 further comprising aggregating opinions of users via classifiers.

17. The computer implemented method of claim 12 further comprising mapping user interaction information to nodes of a social network.

18. The computer implemented method of claim 17 further comprising updating the mapping act.

19. The computer implemented method of claim 12 further comprising indicating relationships as arcs between nodes.

20. A computer implemented system comprising:

at least one processor;

means for implementing a social network, the means connecting a plurality of nodes to represent trust relationships declared by members of the social network for other members of the social network and the means further maintaining recommendations from each of plurality of members of the social network, the recommendations relating to an item; and

means for producing a recommendation on the item for a member represented by a node in the network, the recommendation being based on the recommendations of other members linked to the member through declared trust relationships and on the declared trust relationships, such that different members receive different recommendations.