System and Methods for Updating User Profiles and Providing Selected Documents to Users

• US 20110125767A1
• Filed: 11/25/2009
• Published: 05/26/2011
• Est. Priority Date: 11/25/2009
• Status: Active Grant

• Current Assignee: Telefonaktiebolaget LM Ericsson
• Original Assignee: Telefonaktiebolaget LM Ericsson
• Inventors: Bjork, Jonas, Olsson, Hjalmer, MORITZ, Simon, Carlsson, Richard
• Granted Patent: US 9,251,524 B2
• Time in Patent Office: Days
• US Class Current: 707/754
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0277 Online advertisement

Abstract

The present invention provides an improved recommender system. In some embodiments, the recommender system automatically builds a user profile for a user based solely on the action or actions the user takes with respect to certain documents and the contents of the documents. That is, the profile is based on implicit ratings. In addition to maintaining a user profile for each user, each document in a set of documents will have its own document profile, which is also referred to as a document vector. Advantageously, in some embodiments, the user profile will consist of the same attributes as the document vectors. This advantageous feature enables the recommender system to compare users and documents in order to generate recommendations.

Claims

1. A method performed by a document server for updating and/or creating a profile associated with a user of a client device, wherein the document server comprises a recommender system, the method comprising:

(a) transmitting, from the document server to the user, a document containing a set of one or more terms;

(b) after step (a), receiving, at the document server, information identifying an action or actions the user of the client device took with respect to the document;

(c) in response to receiving the information, determining, for each term included in the set of terms, a value to associate with the term, said value being a function of the user'"'"'s action or actions, thereby establishing a set of term/value pairs, where each said pair consists of one of said terms and its associated value; and

(d) storing the set of term/value pairs in the profile.

2. The method of claim 1, wherein the document contains not only the set of terms but also a second set of terms, and the document is an advertisement.

3. The method of claim 1, wherein the value is a function of the user'"'"'s action or actions and the number of times the term appears in the document.

4. The method of claim 3, wherein

prior to step (c), the profile contains a value associated with a particular term that is included in said set of terms, and

the value determined in step (c) for the particular term is a function of said value contained in the profile (v1), a value (a) corresponding to the user'"'"'s action or actions, and the number of times (n) the term appears in the document.

5. The method of claim 4, wherein the step of determining the value for the particular term comprises computing (v1)+(a)(n).

6. The method of claim 1, wherein the information identifying the user'"'"'s action or actions comprises a value, wherein the value is mapped to a particular action or actions.

7. The method of claim 1, wherein the step of transmitting the document to the user comprises (a) transmitting an e-mail containing the document to an e-mail address associated with the user or (b) transmitting the document directly to the client device.

8. A method performed by a document server for providing a document to a user, comprising:

updating a user vector, the updating step comprising (a) receiving information identifying an action or actions the user took with respect to a particular document that contains a set of one or more terms, (b) in response to receiving the information, computing a value for each term included in the set of terms, each of said values being a function of the user'"'"'s action or actions, and (c) storing the computed values in the user vector such that each said value is associated with its corresponding term;

generating a reduced user vector from the user vector;

using the reduced user vector, select a document from a set of candidate documents; and

transmitting, from the document server to the user, the selected document, wherein

the step of using the reduced user vector to select a document from the set of candidate documents comprises;

obtaining a first reduced document vector representing a first document from the set of candidate documents;

computing a first similarity value representing the similarity between the reduced user vector and the first reduced document vector;

obtaining a second reduced document vector representing a second document from the set of candidate documents;

computing a second similarity value representing the similarity between the reduced user vector and the second reduced document vector; and

comparing the first similarity value with the second similarity value.

9. The method of claim 8, wherein the document contains not only the set of terms but also a second set of terms, and the document is an advertisement.

10. The method of claim 8, wherein each of the plurality of computed values is a function of the user'"'"'s action or actions and the number of times the value'"'"'s corresponding term appears in the document.

11. The method of claim 8, wherein the step of generating a reduced user vector comprises computing xU_k, wherein x is the updated user vector, U_k is a column matrix, and U_k is contains the k eigenvectors of a matrix U.

12. The method of claim 11, further comprising generating the matrix U prior to generating the reduced user vector, wherein the step of generating the matrix U comprises:

generating a matrix M that describes term correlations within all of the candidate documents; and

deriving U from M.

13. The method of claim 8, wherein the step of transmitting the document to the user comprises (a) transmitting an e-mail containing the document to an e-mail address associated with the user or (b) transmitting the document directly to a client device operated by the user.

14. A document serving system for creating and/or updating a user vector associated with a user, comprising:

a data processing system;

a data storage system accessible to the data processing system; and

a network interface coupled to a network, said data processing system being configured to;

(a) transmit to a user, via the network interface and the network, a document containing a set of one or more terms;

(b) receive information identifying an action or actions the user took with respect to the document;

(c) establish a set of term/value pairs in response to receiving the information, where each said pair consists of a term and an associated value, wherein the data processing system is configured to establish the set of term/value pairs by determining, for each term included in the set of terms, a value to associate with the term, said value being a function of the user'"'"'s action or actions; and

(d) store in the data storage system the set of term/value pairs such that the set of term/value pairs is associated with the user.

15. The document serving system of claim 14, wherein the value is a function of the user'"'"'s action or actions and the number of times the term appears in the document.

16. The document serving system of claim 15, wherein

the data storage system stores a value associated with the user and with a particular term that is included in said set of terms, and

the value determined for the particular term is a function of said value stored in the data storage system (v1), a value (a) corresponding to the user'"'"'s action or actions, and the number of times (n) the term appears in the document.

17. The document serving system of claim 16, wherein the data processing system is configured to determine the value for the particular term by computing (v1)+(a)(n).

18. A document serving system for providing documents to a user, comprising:

a data processing system;

a data storage system accessible to the data processing system; and

a network interface coupled to a network, said data processing system being configured to;

update a user vector associated with the user, wherein the data processing system is configured to update the user vector by;

(a) receiving information identifying an action or actions the user took with respect to a particular document that contains a set of one or more terms, (b) computing a value for each term included in the set of terms in response to receiving the information, each of said values being a function of the user'"'"'s action or actions, and (c) storing the computed values in the user vector such that each said value is associated with its corresponding term, thereby updating the user vector;

generate a reduced user vector from the updated user vector;

use the reduced user vector to select a document from a set of candidate documents; and

transmit the selected document to the user via the network interface, wherein

the data processing system is configured to use the reduced user vector to select a document from the set of candidate documents by;

obtaining a first reduced document vector representing a first document from the set of candidate documents;

computing a first similarity value representing the similarity between the reduced user vector and the first reduced document vector;

obtaining a second reduced document vector representing a second document from the set of candidate documents;

computing a second similarity value representing the similarity between the reduced user vector and the second reduced document vector; and

comparing the first similarity value with the second similarity value.

19. The document serving system of claim 18, wherein each of the plurality of computed values is a function of the user'"'"'s action or actions and the number of times the value'"'"'s corresponding term appears in the document.

20. The document serving system of claim 18, wherein the data processing system is configured to generate a reduced user vector by computing xU_k, wherein x is the updated user vector and U_k is a column matrix.

21. The document serving system of claim 20, wherein U_k is contains the k eigenvectors of a matrix U, and the data processing system is further configured to generate the matrix U prior to generating the reduced user vector.

22. The document serving system of claim 21, wherein the data processing system is configured to generate the matrix U by:

generating a matrix M that describes term correlations within all of the candidate documents; and

deriving U from M.

23. A computer program comprising computer readable code means which when executed by a processor causes the processor to perform the steps of:

(a) using a network interface to transmit to a user a document containing a set of one or more terms;

(b) in response to receiving information identifying an action or actions the user took with respect to the document, determining, for each term included in the set of terms, a value to associate with the term, said value being a function of the user'"'"'s action or actions, thereby establishing a set of term/value pairs, where each said pair consists of one of said terms and its associated value; and

(c) storing the set of term/value pairs in a profile associated with the user.

24. The computer program of claim 23, wherein the value is a function of the user'"'"'s action or actions and the number of times the term appears in the document.

25. The computer program of claim 24, wherein

prior to step (b), the profile contains a value associated with a particular term that is included in said set of terms, and

the value determined in step (b) for the particular term is a function of said value contained in the profile (v1), a value (a) corresponding to the user'"'"'s action or actions, and the number of times (n) the term appears in the document.

26. The computer program of claim 25, wherein the step of determining the value for the particular term comprises computing (v1)+(a)(n).

27. A computer program product comprising a computer readable means and a computer program according to claim 26, wherein the computer program is stored in the computer readable means.

28. A computer program comprising computer readable code means which when executed by a processor causes the processor to perform the steps of:

updating a user vector, the updating step comprising (a) receiving information identifying an action or actions the user took with respect to a particular document that contains a set of one or more terms, (b) in response to receiving the information, computing a value for each term included in the set of terms, each of said values being a function of the user'"'"'s action or actions, and (c) storing the computed values in the user vector such that each said value is associated with its corresponding term;

generating a reduced user vector from the user vector;

using the reduced user vector, select a document from a set of candidate documents; and

using a network interface to transmit to the user the selected document, wherein

the step of using the reduced user vector to select a document from the set of candidate documents comprises;

obtaining a first reduced document vector representing a first document from the set of candidate documents;

computing a first similarity value representing the similarity between the reduced user vector and the first reduced document vector;

obtaining a second reduced document vector representing a second document from the set of candidate documents;

computing a second similarity value representing the similarity between the reduced user vector and the second reduced document vector; and

comparing the first similarity value with the second similarity value.

29. A computer program product comprising a computer readable means and a computer program according to claim 28, wherein the computer program is stored in the computer readable means.