SYSTEM AND METHODS FOR PROTECTING THE PRIVACY OF USER INFORMATION IN A RECOMMENDATION SYSTEM

• US 20120246736A1
• Filed: 12/04/2009
• Published: 09/27/2012
• Est. Priority Date: 12/04/2009
• Status: Active Grant

• Current Assignee: Telefonaktiebolaget LM Ericsson
• Original Assignee: Telefonaktiebolaget LM Ericsson
• Inventors: Huang, Vincent, Bjork, Jonas
• Granted Patent: US 8,781,901 B2
• Time in Patent Office: Days
• US Class Current: 726/26
• CPC Class Codes: G06F 21/6254 by anonymising data, e.g. decorrelating personal data from the owner's identification
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0282 Rating or review of business operators or products

Abstract

The invention provides an improved recommender system that includes a client device or service provider server, a trusted function handler module and a recommender module. The recommender system functions to protect the privacy of user rating information maintained by the node (i.e., client device/server) by having the node transform the user rating information using a specific function selected by the function handler and then provide the transformed user rating information to the recommender module. In this way, privacy of the user rating information is maintained because the original user rating information will be unknown to a recommender module.

Claims

1. An apparatus configured to protect the privacy of user information, comprising:

a storage medium storing user rating information;

one or more network interfaces for receiving and transmitting data via a network; and

a data processing system operatively connected to the storage medium and at least one of the one or more network interfaces, the data processing system arranged to;

use one of the network interfaces to receive from a function handler linear transformation information identifying a linear transformation;

apply, to the user rating information, the linear transformation identified by the linear transformation information received from the function handler to produce transformed user rating information, and

use one of the network interfaces to transmit the transformed user rating information towards a recommender.

2. The apparatus of claim 1, wherein the user rating information comprises explicit and/or implicit user information.

3. The apparatus of claim 1, wherein

the data processing system is further arranged to generate a user vector based on the user rating information and further arranged to apply the linear transformation to the user rating information by applying the linear transformation to the user vector, thereby producing a transformed user vector, and

the transformed user rating information comprises the transformed user vector.

4. The apparatus of claim 3, wherein the received linear transformation information identifies one or more of:

a vector rotation, reflection, and scaling.

5. The apparatus of claim 4, wherein the received linear transformation information comprises a transformation matrix corresponding to a vector rotation and the data processing system is arranged to apply the linear transformation to the user vector by multiplying the user vector by the transformation matrix, thereby rotating the user vector according to the corresponding vector rotation.

6. The apparatus of claim 1, wherein the data processing system is further arranged such that, in response to the apparatus receiving from the recommender recommendation information, the data processing system is arranged to apply to the recommendation information an inverse of the linear transformation identified by the received linear transformation information.

7. The apparatus of claim 6, wherein the recommendation information comprises a recommendation vector and the step of applying to the recommendation information the inverse of the linear transformation comprises rotating the recommendation vector.

8. A function handler apparatus for use in protecting the privacy of user information, comprising:

one or more network interfaces for receiving data via a network and transmitting data via the network; and

a data processing system operatively connected to the network interface, the data processing system arranged to;

use one of the network interfaces to transmit first linear transformation information identifying a first linear transformation to a first node;

use one of the network interfaces to transmit second linear transformation information identifying a second linear transformation different from the first linear transformation to a second node; and

use one of the network interfaces to transmit to a recommender information identifying a difference between the first linear transformation and the second linear transformation.

9. The function handler apparatus of claim 8, wherein the first linear transformation information identifies a first vector rotation, and the second linear transformation information identifies a second vector rotation, and the information identifying the difference between the first linear transformation and the second linear transformation identifies the difference between the first vector rotation and the second vector rotation.

10. The function handler apparatus of claim 8, wherein the first node belongs to a first service provider and the second node belongs to a second service provider.

11. The function handler apparatus of claim 8, wherein the data processing system is further arranged such that, in response to the function handler receiving a request from a node, the data processing system selects a linear transformation and uses one of the network interfaces to transmit to the node linear transformation information identifying the selected linear transformation.

12. The function handler apparatus of claim 11, wherein

the step of selecting a linear transformation comprises selecting a transformation matrix;

and

the data processing system is arranged to select the transformation matrix using one or more parameters included in the request.

13. The function handler apparatus of claim 13, wherein the one or more parameters include one or more of:

a recommendation accuracy parameter and a privacy level parameter.

14. A recommender apparatus for use in protecting the privacy of user information, comprising:

a network interface operable to receive data via a network and transmit data via the network; and

a data processing system operatively connected to the network interface, the data processing system arranged to;

(i) transform a first transformed user vector received from a first node using difference information received from a function handler to produce a further transformed user vector, (ii) perform a user similarity procedure by comparing a second transformed user vector received from a second node with the further transformed user vector, and (iii) use the network interface to provide a result of the user similarity procedure to the first node and/or the second node, wherein

the difference information identifies a difference between a first linear transformation used to create the first transformed user vector and a second linear transformation used to create the second transformed user vector, with the second linear transformation being different than the first linear transformation.

15. The recommender apparatus of claim 14, wherein

the first linear transformation is a first vector rotation,

the second linear transformation is a second vector rotation, and

the difference information identifying the difference between the first linear transformation and the second linear transformation comprises a transformation matrix corresponding to the difference between the first vector rotation and the second vector rotation.

16. The recommender apparatus of claim 15, wherein the data processing system is configured to transform the first transformed user vector using the difference information by multiplying the first transformed user vector by the transformation matrix.

17. The recommender apparatus of claim 14, wherein the first node belongs to a first service provider and the second node belongs to a second service provider.

18. The recommender apparatus of claim 14, wherein the data processing system is further arranged to:

use the further transformed user vector to select an item to recommend to the user associated with the first transformed user vector, and

use a network interface to transmit to the first node information pertaining to the selected item.

19. A method for protecting the privacy of user information in a system comprising a node, a function handler, and a recommender, comprising:

storing user rating information in a storage medium accessible to the node;

transmitting, from the node to the function handler, a request;

receiving the request at the function handler;

in response to receiving the request, selecting a linear transformation and transmitting to the node linear transformation information identifying the selected linear transformation;

receiving at the node the linear transformation information transmitted by the function handler;

applying the linear transformation identified by the received linear transformation information to the user rating information to produce transformed user rating information;

transmitting, from the node to the recommender, the transformed user rating information;

receiving the transformed user rating information at the recommender;

using the transformed user rating information to perform a similarity procedure; and

transmitting, from the recommender to the node, a result of the similarity procedure.

20. The method of claim 19, wherein

the user rating information comprises a user vector associated with explicit and/or implicit user preference information, and

the linear transformation information transmitted to the node by the function handler identifies one or more of:

a vector rotation, reflection, and scaling.

21. The method of claim 19, wherein

the user rating information comprises a user vector associated with explicit and/or implicit user preference information, and

the step of applying the linear transformation to the user rating information comprises multiplying the user vector by a transformation matrix identified by the linear transformation information transmitted to the node by the function handler.

22. The method of claim 19, wherein the system further comprises a second node and the method further comprises:

storing second user rating information in a storage medium accessible to the second node;

transmitting, from the second node to the function handler, a second request;

receiving the second request at the function handler;

in response to receiving the second request, selecting a second linear transformation and transmitting to the second node second linear transformation information identifying the second linear transformation;

receiving at the second node the second linear transformation information transmitted by the function handler;

applying the second linear transformation identified by the received second linear transformation information to the second user rating information to produce second transformed user rating information;

transmitting, from the second node to the recommender, the second transformed user rating information;

receiving, at the recommender, the second transformed user rating information;

transmitting, from the function handler to the recommender, difference information identifying a difference between the first linear transformation and the second linear transformation;

receiving, at the recommender, the difference information transmitted from the function handler;

transforming the second transformed user rating information using the difference information to produce further transformed user rating information;

performing, at the recommender, a user similarity procedure using the first transformed user rating information and the further transformed user rating information; and

transmitting, from the recommender to the first node and/or the second node, a result of the user similarity procedure.