RECOMMENDATION SYSTEM WITH MULTIPLE INTEGRATED RECOMMENDERS

• US 20090006398A1
• Filed: 06/29/2007
• Published: 01/01/2009
• Est. Priority Date: 06/29/2007
• Status: Active Grant

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Kim, Sung H., Lam, Shing Yan, Ionkov, George M., Witt, Brett W., Chakrabarti, Kushal
• Granted Patent: US 8,260,787 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: H04L 51/02 using automatic reactions or user delegation, e.g. automatic replies or chatbot-generated messages

Abstract

A recommendations system is provided in various embodiments for selecting items to recommend to a user. The system includes a recommendation engine with a plurality of recommenders, and each recommender identifies a different type of reason for recommending items. In one embodiment, each recommender retrieves item preference data and generates candidate recommendations responsive to a subset of that data. The recommenders also score the candidate recommendations. In certain embodiments, a normalization engine normalizes the scores of the candidate recommendations provided by each recommender. A candidate selector selects at least a portion of the candidate recommendations based on the normalized scores to provide as recommendations to the user. The candidate selector also outputs the recommendations with associated reasons for recommending the items.

Claims

1. A computer-implemented method of normalizing item recommendation scores, the method comprising:

receiving scores for candidate recommendations from first and second recommenders configured to provide recommendations to a target user, the first recommender configured to assign the scores to the candidate recommendations using a different scoring scale from the second recommender;

for each recommender, normalizing the scores assigned by the recommender by;

calculating a range of scores, the range comprising a difference between a minimum score and a maximum score, and

calculating normalized scores as a function of the range; and

using the normalized scores to select at least a portion of the candidate recommendations to recommend to the target user.

2. The method of claim 1, further comprising translating the scores prior to calculating the scores as a function of the range.

3. The method of claim 1, wherein calculating normalized scores as a function of the range comprises dividing the scores by the range.

4. The method of claim 1, wherein normalizing the scores assigned by the recommender further comprises determining a new range for the normalized scores.

5. The method of claim 4, wherein calculating normalized scores as a function of the range comprises mapping the scores from the range to the new range by applying a transformation to the scores.

6. The method of claim 1, wherein calculating a range of scores comprises using a window to calculate a new minimum score if a score received from a recommender is less than the minimum score and calculating a new maximum score if a score received from a recommender is greater than the maximum score.

7. The method of claim 6, further comprising resetting the window.

8. The method of claim 1, wherein normalizing the scores assigned by the recommender further comprises:

normalizing the scores using a plurality of computing devices; and

reducing a number of digits of precision in the scores, such that variability in the normalized scores among the computing devices is reduced.

9. The method of claim 1, further comprising:

partitioning the scores into a set of positive scores corresponding to positive candidate recommendations and a set of negative scores corresponding to negative candidate recommendations; and

normalizing the positive scores separately from the negative scores.

10. The method of claim 9, further comprising combining normalized positive scores with normalized negative scores to obtain overall normalized scores for the candidate recommendations.

11. The method of claim 1, further comprising applying exponential decay to scores for equivalent items provided by both the first and second recommenders.

12. The method of claim 1, further comprising assigning weights to candidate recommendations received from the first and second recommenders.

13. The method of claim 12, further comprising adjusting the weights based on direct or indirect feedback received from the user.

14. A computer-implemented method of normalizing item recommendation scores, the method comprising:

receiving scores for candidate recommendations from first and second recommenders configured to provide recommendations to a target user, the first recommender configured to assign the scores to the candidate recommendations using a different scoring scale from the second recommender;

for each recommender, normalizing the scores assigned by the recommender by;

combining the scores for at least some of the candidate recommendations to generate a combined score, and

calculating normalized scores as a function of the combined score and the scores for at least some of the candidate recommendations; and

using the normalized scores to select at least a portion of the candidate recommendations to recommend to the target user.

15. The method of claim 14, further comprising translating the scores by a difference between a minimum score and a desired normalized minimum score in response to determining that the minimum score differs from the desired normalized minimum score.

16. The method of claim 14, further comprising adjusting the number of scores combined in the combined score to reduce the impact of outliers on the normalized scores.

17. The method of claim 14, wherein combining the scores for at least some of the candidate recommendations comprises summing the scores.

18. The method of claim 14, further comprising assigning weights to candidate recommendations received from the first and second recommenders.

19. A computer-implemented method of normalizing item recommendation scores, the method comprising:

receiving scores for candidate recommendations from first and second recommenders configured to provide recommendations to a target user, the first recommender configured to assign the scores to the candidate recommendations using a different scoring scale from the second recommender;

for each recommender, normalizing the scores assigned by the recommender by assigning percentile rankings to the scores and using the percentile rankings as normalized scores; and

using the normalized scores to select at least a portion of the candidate recommendations to recommend to the target user.

20. The method of claim 19, wherein assigning percentile rankings to the scores comprises dynamically assigning percentile rankings to the scores as the scores are received from the recommenders.

21. The method of claim 19, further comprising:

partitioning the scores into a set of positive scores corresponding to positive candidate recommendations and a set of negative scores corresponding to negative candidate recommendations; and

normalizing the positive scores separately from the negative scores.

22. The method of claim 21, further comprising reversing the rankings of the negative scores.

23. A system for normalizing item recommendation scores, the system comprising:

a plurality of recommenders configured to assign scores to candidate recommendations using different scoring scales;

a normalization engine operative to normalize scores assigned by the plurality of recommenders, the normalization engine configured to;

calculate a range of scores, the range comprising a difference between a minimum score and a maximum score, and

calculate normalized scores as a function of the range; and

a candidate selector configured to use the normalized scores to select at least a portion of the candidate recommendations to recommend to a target user.

24. A system for normalizing item recommendation scores, the system comprising:

a plurality of recommenders configured to assign scores to candidate recommendations using different scoring scales;

a normalization engine operative to normalize scores assigned by the plurality of recommenders, the normalization engine configured to;

combine the scores for at least some of the candidate recommendations to generate a combined score, and

calculate normalized scores as a function of the combined score and of the scores for at least some of the candidate recommendations; and

a candidate selector configured to use the normalized scores to select at least a portion of the candidate recommendations to recommend to a target user.

25. A system for normalizing item recommendation scores, the system comprising:

a plurality of recommenders configured to assign scores to candidate recommendations using different scoring scales;

a normalization engine operative to normalize scores assigned by the plurality of recommenders, the normalization engine configured to assign percentile rankings to the scores and use the percentile rankings as normalized scores; and

a candidate selector configured to use the normalized scores to select at least a portion of the candidate recommendations to recommend to a target user.