System for delivering recommendations

• US 7,013,238 B1
• Filed: 07/15/2003
• Issued: 03/14/2006
• Est. Priority Date: 02/24/2003
• Status: Expired due to Fees

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Weare, Christopher B.
• Primary Examiner(s): Hoff, Marc S.
• Assistant Examiner(s): SUAREZ, FELIX E
• Application Number:
  US10/621,268
• Time in Patent Office: 973 Days
• Field of Search: 702/179, 702181-183, 702/186, 702/188, 702/189, 702/194, 235/385, 705/7, 709/203
• US Class Current: 702/182
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising

Abstract

A system for generating recommendations which automatically optimizes over time without human intervention is disclosed. While known recommendation systems tended to be either attribute based or collaborative filtering based, the present system adjusts its internal parameters in response to the measured performance of the recommendations as determined by user behavior. In one embodiment, the user provides feedback in the form of ratings for the recommendations. This feedback is processed and then utilized to adjust the internal parameters of the system. Over time, the system settles on optimal parameter values.

Claims

1. In a computer system, a method for providing recommendations of items to a user, the method comprising:

evaluating the items in accordance with sets of parameters;

recommending selected items to a user based on the evaluation of the items;

receiving from the user feedback regarding the recommendations;

adjusting the sets of parameters based on the feedback, wherein a genetic algorithm is utilized for adjusting the sets of parameters, and a fitness evaluation stage is utilized during which the fitness of each parameter set is evaluated and the results of the fitness evaluation are further utilized in a survival stage which determines the survival of each of the parameter sets.

2. The method of claim 1, wherein the feedback from the user is in the form of user ratings of the recommendations.

3. The method of claim 1, wherein the evaluation of the items includes determining affinity values for the items.

4. The method of claim 1, wherein other items that have been selected by the user as favorites are utilized as references for the evaluation of the items.

5. The method of claim 1, wherein the results of the fitness evaluation are further utilized in a mating stage which determines the mating of the parameter sets.

6. The method of claim 5, wherein when two parameter sets mate, a resulting parameter set contains elements from each of the original parameter sets.

7. The method of claim 1, wherein the number of surviving parameter sets is generally limited to 50 or fewer parameter sets.

8. The method of claim 1, wherein an initial population of parameter sets is created using known values.

9. The method of claim 1, wherein an initial population of parameter sets is created through a random process.

10. The method of claim 1, wherein a mutation stage is utilized during which mutated parameter sets are created.

11. In a computer system, a method for providing recommendations of items to a user, the method comprising:

evaluating the items in accordance with sets of parameters;

recommending selected items to a user based on the evaluation of the items;

receiving from the user feedback regarding the recommendations;

adjusting the sets of parameters based on the feedback, wherein a genetic algorithm is utilized for adjusting the sets of parameters and floating point genes are utilized by the genetic algorithm.

12. The method of claim 11, wherein the feedback from the user is in the form of user ratings of the recommendations.

13. The method of claim 11, wherein the evaluation of the items includes determining affinity values for the items.

14. The method of claim 11, wherein other items that have been selected by the user as favorites are utilized as references for the evaluation of the items.

15. The method of claim 11, wherein a fitness evaluation stage is utilized during which the fitness of each parameter set is evaluated.

16. The method of claim 15, wherein the results of the fitness evaluation are further utilized in a mating stage which determines the mating of the parameter sets.

17. The method of claim 16, wherein when two parameter sets mate, a resulting parameter set contains elements from each of the original parameter sets.

18. The method of claim 11, wherein a mutation stage is utilized during which mutated parameter sets are created.

19. In a computer system, a method for providing recommendations of items to a user, the method comprising:

evaluating the items in accordance with sets of parameters;

recommending selected items to a user based on the evaluation of the items;

receiving from the user feedback regarding the recommendations;

adjusting the sets of parameters based on the feedback, wherein a genetic algorithm is utilized for adjusting the sets of parameters and binary genes are utilized by the genetic algorithm.

20. The method of claim 19, wherein the feedback from the user is in the form of user ratings of the recommendations.

21. The method of claim 19, wherein the evaluation of the items includes determining affinity values for the items.

22. The method of claim 19, wherein other items that have been selected by the user as favorites are utilized as references for the evaluation of the items.

23. The method of claim 19, wherein a fitness evaluation stage is utilized during which the fitness of each parameter set is evaluated.

24. The method of claim 23, wherein the results of the fitness evaluation are further utilized in a mating stage which determines the mating of the parameter sets.

25. The method of claim 24, wherein when two parameter sets mate, a resulting parameter set contains elements from each of the original parameter sets.

26. The method of claim 19, wherein a mutation stage is utilized during which mutated parameter sets are created.

27. A system for providing recommendations to a user, the system comprising:

an affinity predictor for receiving information regarding items and determining affinity values for the items, the affinity values being determined according to sets of parameters, the affinity values being used to provide recommendations to the user;

a parameter controller for adjusting the sets of parameters in accordance with feedback from the user regarding the recommendations, wherein the parameter controller utilizes a genetic algorithm; and

an executing component for executing parameter sets that are determined to be undesirable.

28. The system of claim 27, further comprising a mating component for determining the mating of parameter sets.

29. The system of claim 27, further comprising a performance monitor which harvests recommendation performance information from the user.

30. The system of claim 27, further comprising a recommendation factory which selects items that have the highest expected affinity levels as determined by the affinity predictor.

31. The system of claim 30, further comprising a recommendation table which the recommendation factory writes the recommendations to, the recommendations being loaded for the user from the recommendation table.

32. The system of claim 27, further comprising a mutating component for creating mutated parameter sets.