Using cross-site relationships to generate recommendations

• US 20070208729A1
• Filed: 03/06/2006
• Published: 09/06/2007
• Est. Priority Date: 03/06/2006
• Status: Active Grant

• Current Assignee: Aggregate Knowledge Incorporated (Neustar Incorporated)
• Original Assignee: Aggregate Knowledge Incorporated (Neustar Incorporated)
• Inventors: Martino, Paul
• Granted Patent: US 7,788,358 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/337 Profile generation, learning or modification
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06Q 30/0255 based on user history
• CPC Class Codes: G06Q 30/0633 Lists, e.g. purchase orders, compilation or processing

Abstract

A relationship server tracks end-user interactions across multiple web sites and generates recommendations. The web sites observe relationships established by end-user interactions. If end-users provide the same personally identifiable information to multiple web sites, the sites generate the same unique identifier for those end-users. The web sites send messages to the relationship server that reference the end-users using the identifiers and describe the relationships observed for the end-users. The relationship server receives messages from multiple web sites and canonicalizes them to produce an efficient representation of the relationships. Upon receiving a message requesting a recommendation based on an item, the relationship server performs collaborative filtering using the relationship data to identify a list of items to recommend. The relationship server sends the recommendations to the requesting entity and the recommendations are presented to the end-user.

Claims

1. A system for generating recommendations, comprising:

a communications module for receiving messages describing relationships from a plurality of computers via a network;

a relationship storage module for storing canonicalized representations of the relationships described by the messages in a memory; and

a recommendation generation module for generating recommendations based on the canonicalized representations of the relationships stored in the memory.

2. The system of claim 1, wherein a message describing a relationship received by the communications module includes a label identifying an item and a predicate describing a type of the item.

3. The system of claim 1, wherein a message describing a relationship received by the communications module includes an intent describing an action that established the relationship.

4. The system of claim 1, further comprising:

a normalization module for normalizing the messages describing relationships received from the plurality of computers.

5. The system of claim 1, further comprising:

a canonicalization module for generating canonicalized representations of the messages optimized for efficient storage in the memory.

6. The system of claim 5, wherein a message describing a relationship includes a label identifying an item and a predicate describing a type of the item and wherein the canonicalization module substitutes integer values for the item and predicate to form the canonicalized representation.

7. The system of claim 5, wherein a message describing a relationship includes a label identifying an item and a predicate describing a type of the item and wherein the canonicalization module generates a canonicalized representation of the item and predicate that fits into a single memory word of a computer.

8. The system of claim 1, wherein the recommendation generation module performs collaborative filtering using the canonicalized representations of the relationships to generate the recommendations.

9. The system of claim 1, wherein the communications module is adapted to receive a message requesting recommendations, the message specifying a label identifying an item and a predicate describing a type of the item, and wherein the recommendation generation module performs collaborative filtering to identify a set of items related to the item identified in the message requesting recommendations.

10. The system of claim 1, wherein the communications module is adapted to receive a message requesting recommendations based on an item, the message specifying at least one constraint on items to be returned as recommendations, and wherein the recommendation generation module performs collaborative filtering to identify a set of items related to the item identified in the message requesting recommendations and satisfying the at least one constraint.

11. A system for providing recommendations of items to an end-user comprising:

a relationship observation module for observing relationships established by the end-user through interactions with a web site;

a communications module for sending messages describing the observed relationships to a relationship server via a computer network and for receiving from the relationship server recommendations for the end-user based at least in part on observed relationships described by messages sent by a plurality of web sites; and

a recommendations module for presenting the received recommendations to the end-user.

12. The system of claim 11, further comprising:

an interaction module for providing web pages via the computer network to a client utilized by the end-user to enable the end-user to interact with the web site.

13. The system of claim 11, further comprising:

an identification module for identifying the end-user and associating the end-user'"'"'s interactions.

14. The system of claim 13, wherein the identification module stores a cookie containing a value identifying the end-user at a client used by the end-user to access the web site.

15. The system of claim 14, wherein the value contained by the cookie is derived from personally identifiable information provided by the end-user, the plurality of web sites use the value to identify the end-user, and the value is included with messages describing observed relationships sent to the relationship server.

16. The system of claim 11, wherein the communications module is adapted to send a message describing a relationship that includes a first label identifying a first item, a first predicate describing a type of the first item, a second label identifying a second item, and a second predicate describing a type of the second item, and wherein the message indicates that the first and second items are related.

17. The system of claim 16, wherein the message describing the relationship indicates that the first and second labels identify a same item.

18. The system of claim 11, wherein the communications module is adapted to send a message describing a relationship that includes an intent describing an action that established the relationship.

19. The system of claim 11, wherein the communications module is adapted to send a message requesting recommendations to the relationship server via the computer network, the message requesting recommendations including a label identifying an item and a predicate describing a type of the item, and wherein the recommendations received from the relationship server are based at least in part on the label and predicate.

20. A system for providing recommendations, comprising:

a plurality of web sites coupled to a network, the web sites interacting with clients and observing relationships established by the clients through the interactions, generating messages describing the relationships, and receiving messages describing recommendations for particular clients responsive at least in part to the described relationships; and

a relationship server coupled to the network, receiving the messages describing the relationships from the plurality of web sites, generating the messages describing recommendations for the particular clients responsive at least in part to the relationships described by the messages, and providing the messages describing the recommendations to the web sites.

21. The system of claim 20, wherein an end-user of a client provides personally identifiable information to the plurality of web sites, the plurality of web sites generate a same identifier from the personally identifiable information, and the plurality of web sites use the same identifier in messages to the relationship server describing relationships associated with the end-user.

22. The system of claim 20, wherein the relationship server performs collaborative filtering using the relationships described by the messages from the plurality of web sites to generate the messages describing the recommendations.

23. The system of claim 20, further comprising:

an aggregation module coupled to the network and remote from the clients, the aggregation module receiving identifiers of end-users of the clients, interacting with the relationship server using the identifiers to obtain recommendations for the end-users of the clients, and sending the recommendations to the clients.

24. A method of providing recommendations, comprising:

tracking end-user interactions across a plurality of unrelated web sites;

establishing a set of relationships for the end-user responsive to the tracked interactions;

generating recommendations of items for the end-user responsive at least in part to the established set of relationships; and

providing the recommendations to the end-user.

25. The method of claim 24, wherein tracking end-user interactions comprises:

receiving personally identifiable information from the end-user;

generating a unique identifier for the end-user responsive to the personally identifiable information, wherein the plurality of unrelated web sites generate the same unique identifier responsive to the same personally identifiable information; and

associating the unique identifier with the end-user interactions.

26. The method of claim 24, wherein generating recommendations comprises:

performing collaborative filtering responsive at least in part to the set of relationships for the end-user and sets of relationships established for other end-users through interactions across the plurality of unrelated web sites.

27. A computer program product having a computer-readable medium having computer program instructions embodied therein for providing recommendations of items to an end-user, comprising:

a communications module for sending messages describing relationships established by an end-user through interactions with a web site to a relationship server via a computer network and for receiving from the relationship server recommendations for the end-user based at least in part on observed relationships described by messages sent to the relationship server by a plurality of web sites; and

a recommendations module for presenting the received recommendations to the end-user.

28. The computer program product of claim 27, further comprising:

a relationship observation module for observing relationships established by the end-user through interactions with the web site.

29. The computer program product of claim 27, wherein the recommendations module is further adapted to provide recommendations to a plurality of end-users of a plurality of remote clients via the computer network.

30. The computer program product of claim 27, further comprising:

an interaction module for providing web pages via the computer network to a client utilized by the end-user to enable the end-user to interact with the web site.

31. The computer program product of claim 27, further comprising:

an identification module for identifying the end-user and associating the end-user'"'"'s interactions.

32. The computer program product of claim 31, wherein the identification module stores a cookie containing a value identifying the end-user at a client used by the end-user to access the web site.

33. The computer program product of claim 32, wherein the value contained by the cookie is derived from personally identifiable information provided by the end-user, the plurality of web sites use the value to identify the end-user, and the value is included with messages describing observed relationships sent to the relationship server.

34. The computer program product of claim 27, wherein the communications module is adapted to send a message requesting recommendations to the relationship server via the computer network, the message requesting recommendations including a label identifying an item and a predicate describing a type of the item, and wherein the recommendations received from the relationship server are based at least in part on the label and predicate.