Engagement-Oriented Recommendation Principle

• US 20090055257A1
• Filed: 08/24/2007
• Published: 02/26/2009
• Est. Priority Date: 08/24/2007
• Status: Active Grant

• Current Assignee: R2 Solutions LLC (Acacia Research Corporation)
• Original Assignee: Yahoo! Inc. (Apollo Global Management, Inc.)
• Inventors: Chien, Shu-Yao, Seth, Amitabh
• Granted Patent: US 7,904,303 B2
• Time in Patent Office: Days
• US Class Current: 705/14.690
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0273 Determination of fees for advertising

Abstract

A method selects a predictor item that has a relevance to a user. The method receives a set of affinity items having affinity scores that relate the predictor item to the affinity items. The method filters the list of affinity items based on the affinity scores, and selects a first set of affinity items from the filtered items. For each selected affinity item, the method calculates a difference score from the predictor item, and selects a first affinity item based on the difference score for the first affinity item. Preferably, content is presented to the user based on the selected first affinity item. Additional embodiments include a system and/or computer readable medium having instructions for execution of the foregoing.

Claims

1. A method of targeting, the method comprising:

selecting a predictor item having a relevance to a user;

receiving a plurality of affinity items having affinity scores that relate the predictor item to the affinity items, thereby generating a plurality of received affinity items;

filtering the received affinity items based on the affinity scores;

selecting from the received affinity items a first set of affinity items;

for each selected affinity item in the first set, calculating a difference score measuring a difference of the selected affinity item from the predictor item; and

from the first set, selecting a first affinity item based on the difference score for the first affinity item.

2. The method of claim 1, further comprising:

presenting to the user, content based on the selected first affinity item.

3. The method of claim 1, the filtering comprising identifying affinity items having affinity scores that are greater than a threshold.

4. The method of claim 1, wherein the predictor item and the affinity items are categorized in a hierarchical tree structure.

5. The method of claim 1, wherein the difference score is based on a distance metric between the predictor item and the first affinity item.

6. The method of claim 1, further comprising:

for the first affinity item, determining a first total score comprising the affinity score and the difference score for the first affinity item;

for a second affinity item, determining a second total score comprising the affinity score and the difference score for the second affinity item; and

selecting the first affinity item further comprising;

comparing the first and second total scores for the first and second affinity items.

7. The method of claim 6, the step of determining a total score further comprising:

applying a weight factor to the difference score thereby scaling the difference score to the affinity score, and further thereby adjusting the significance of the difference score in the total score.

8. A computer readable medium storing a program having sets of instructions for targeting, the sets of instructions comprising instructions for:

selecting a predictor item having a relevance to a user;

receiving a set of affinity items having affinity scores that relate the predictor item to the affinity items;

filtering the list of affinity items based on the affinity scores;

selecting a first set of affinity items;

for each selected affinity item, calculating a difference score from the predictor item;

selecting a first affinity item based on the difference score for the first affinity item.

9. The computer readable medium of claim 8, further comprising instructions for:

presenting to the user, content based on the selected first affinity item.

10. The computer readable medium of claim 8, the filtering comprising identifying affinity items having affinity scores that are greater than a threshold.

11. The computer readable medium of claim 8, wherein the predictor item and the affinity items are categorized in a hierarchical tree structure.

12. The computer readable medium of claim 8, wherein the difference score is based on a distance metric between the predictor item and the first affinity item.

13. The computer readable medium of claim 8, further comprising instructions for:

for the first affinity item, determining a first total score comprising the affinity score and the difference score for the first affinity item;

for a second affinity item, determining a second total score comprising the affinity score and the difference score for the second affinity item; and

the instructions for selecting the first affinity item further comprising instructions for;

comparing the first and second total scores for the first affinity item and the second affinity item.

14. The computer readable medium of claim 13, determining a total score further comprising:

applying a weight factor to the difference score thereby scaling the difference score to the affinity score, and further thereby adjusting the significance of the difference score in the total score.

15. A system for targeting, the system configured for:

selecting a predictor item having a relevance to a user;

receiving a set of affinity items having affinity scores that relate the predictor item to the affinity items;

filtering the list of affinity items based on the affinity scores;

selecting a first set of affinity items;

for each selected affinity item, calculating a difference score from the predictor item; and

selecting a first affinity item based on the difference score for the first affinity item.

16. The system of claim 15, further configured for:

presenting to the user, content based on the selected first affinity item.

17. The system of claim 15, further configured for:

identifying affinity items having affinity scores that are greater than a threshold.

18. The system of claim 15, wherein the predictor item and the affinity items are categorized in a hierarchical tree structure.

19. The system of claim 15, wherein the difference score is based on a distance metric between the predictor item and the first affinity item.

20. The system of claim 15, further configured for:

for the first affinity item, determining a first total score comprising the affinity score and the difference score for the first affinity item;

for a second affinity item, determining a second total score comprising the affinity score and the difference score for the second affinity item; and

the system comprising a comparator for comparing the first and second total scores for the first affinity item and the second affinity item.

21. The system of claim 20, determining a total score further comprising:

a weight factor for scaling the difference score to the affinity score, and further for adjusting the significance of the difference score in the total score.