System and method for neighborhood optimization for content recommendation

• US 20080031280A1
• Filed: 11/21/2006
• Published: 02/07/2008
• Est. Priority Date: 08/01/2006
• Status: Active Grant

• Current Assignee: Sony Corporation (Sony Group Corp.), Sony Electronics Inc. (Sony Group Corp.)
• Original Assignee: Sony Corporation (Sony Group Corp.), Sony Electronics Inc. (Sony Group Corp.)
• Inventors: Lin, Frank Li-De, Hwang, Paul Jin, Georgis, Nikolaos
• Granted Patent: US 7,559,072 B2
• Time in Patent Office: Days
• US Class Current: 370/486
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: H04H 60/66 for using the result on distributors' side
• CPC Class Codes: H04N 21/252 Processing of multiple end-users' preferences to derive collaborative data
• CPC Class Codes: H04N 21/4661 Deriving a combined profile for a plurality of end-users of the same client, e.g. for family members within a home user profiles in network data switching protocols H04L67/306
• CPC Class Codes: H04N 7/163 by receiver means only

Abstract

A cost function is stochastically optimized using, e.g., simulated annealing to render a neighborhood of entities based on which content recommendations can be provided to a user of a home entertainment system. The cost function represents a normalized sum of rating similarity scores from entities of the neighborhood that are related to content items viewed by the user.

Claims

1. A method for recommending content to a home system for display of the content on a display device of the home system, comprising:

defining a neighborhood of other users using the cost function;

$C=f<br><br>\left(N\right)=\frac{1}{<br><br>N<br><br>}<br><br>\underset{s\in <br><br>\hat{S}}{\sum <br><br>}<br><br>\left(\frac{1}{\underset{c^{\mathrm{\prime <br><br>}}\in <br><br>\hat{C}}{\sum <br><br>}<br><br><br><br>\mathrm{sim}<br><br>\left(c,c^{\mathrm{\prime <br><br>}}\right)<br><br>}\times <br><br>\underset{c^{\mathrm{\prime <br><br>}}\in <br><br>\hat{C}}{\sum <br><br>}<br><br>\mathrm{sim}<br><br>\left(c,c^{\mathrm{\prime <br><br>}}\right)\times <br><br>r_{c^{\mathrm{\prime <br><br>}},s}\right)$

based on the neighborhood, providing at least one content recommendation to a user of the home system.

2. The method of claim 1, wherein one term of the cost function represents a sum of rating similarity scores from entities of the neighborhood and related to content items viewed by the user.

3. The method of claim 1, wherein the cost function is normalized.

4. The method of claim 2, wherein the rating similarity scores are cosine-based similarity scores.

5. The method of claim 1, wherein the cost function is optimized using a stochastic method.

6. The method of claim 5, wherein the stochastic method is simulated annealing.

7. The method of claim 6, wherein the simulated annealing is undertaken using at least two operators selected from the group consisting of moving a non-neighborhood entity into the neighborhood (“

move”

), moving a neighborhood entity out of the neighborhood (“

move”

), swapping an entity who is not in the neighborhood with an entity who is in the neighborhood (“

swap”

), swapping an entity who is not in the neighborhood with an entity who is in the neighborhood twice in succession (“

swap-swap”

), and executing a swap then a move in immediate succession.

8. A system receiving ratings from entities that collectively comprise potential neighbors of a subject user in a neighborhood and returning content recommendations to the subject user at least partially based thereon, comprising:

at least one server programmed to stochastically establish a pseudo-optimum neighborhood based at least in part on the ratings, and to return content recommendations at least partially based on the neighborhood.

9. The system of claim 8, wherein the server defines the neighborhood using a cost function, wherein one term of the cost function represents a sum of rating similarity scores from entities of the neighborhood and related to content items viewed by the user.

10. The system of claim 9, wherein the cost function is normalized.

11. The system of claim 10, wherein the rating similarity scores are cosine-based similarity scores.

12. The system of claim 9, wherein the cost function is optimized using a stochastic method.

13. The system of claim 12, wherein the stochastic method is simulated annealing.

14. The system of claim 13, wherein the simulated annealing is undertaken using at least two operators selected from the group consisting of moving a non-neighborhood entity into the neighborhood (“

move”

), moving a neighborhood entity out of the neighborhood (“

move”

), swapping an entity who is not in the neighborhood with an entity who is in the neighborhood (“

swap”

), swapping an entity who is not in the neighborhood with an entity who is in the neighborhood twice in succession (“

swap-swap”

), and executing a swap then a move in immediate succession.

15. A computer iteratively computing a cost function representing a sum of rating similarity scores from entities of a neighborhood of a home entertainment system and related to content items viewed by a user of the home entertainment system, the neighborhood being used to recommend content to the user of the home entertainment system.

16. The computer of claim 15, wherein the cost function is normalized.

17. The computer of claim 16, wherein the rating similarity scores are cosine-based similarity scores.

18. The computer of claim 15, wherein the cost function is optimized using a stochastic method.

19. The computer of claim 18, wherein the stochastic method is simulated annealing.

20. The computer of claim 19, wherein the simulated annealing is undertaken using at least two operators selected from the group consisting of moving a non-neighborhood entity into the neighborhood (“

move”

), moving a neighborhood entity out of the neighborhood (“

move”

), swapping an entity who is not in the neighborhood with an entity who is in the neighborhood (“

swap”

), swapping an entity who is not in the neighborhood with an entity who is in the neighborhood twice in succession (“

swap-swap”

), and executing a swap then a move in immediate succession.