Entropy-based mixing and personalization

• US 9,015,170 B2
• Filed: 09/06/2013
• Issued: 04/21/2015
• Est. Priority Date: 07/07/2009
• Status: Active Grant

• Current Assignee: R2 Solutions LLC (Acacia Research Corporation)
• Original Assignee: Yahoo! Inc. (Apollo Global Management, Inc.)
• Inventors: Xu, Zhichen, Skarra-Gallagher, Aurora, Sample, Neal, Fu, Yun
• Primary Examiner(s): GOFMAN, ALEX N
• Application Number:
  US14/019,931
• Publication Number:
  US 20140006422A1
• Time in Patent Office: 592 Days
• Field of Search: 707/748, 707/749
• US Class Current: 707/748
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/90335 Query processing
• CPC Class Codes: G06Q 10/10 Office automation; Time management
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 50/01 Social networking

Abstract

Techniques are provided for selecting a diverse mix of content items that may be displayed to a user. Content items such as user-generated events are received from a variety of sources. One or more content items are added to a set of content items based on a diversity of characteristics. The diversity of characteristics for the one or more content items may be calculated by measuring a diversity of characteristics of the set as if the one or more content items were added to the set. Content items that produce a greater diversity are selected for addition to the set. The set is displayed to the user, who is provided with a more meaningful mix of content due to the greater diversity in content.

Claims

1. A method in a computer system for selecting a content item to present to a user, the method comprising:

obtaining a set of a plurality of content items already presented to the user;

removing at least one content item from the set of content items already presented to the user to select a subset of one or more already presented content items to be presented again to the user;

receiving a plurality of content items not yet presented to the user, each generated at one of a plurality of different sources;

classifying each of the received content items into one or more of a plurality of groups of content items based at least on the source of the plurality of sources at which the content item is generated;

selecting a candidate set of a plurality of content items from the groups of classified content items;

for each content item in the candidate set of content items, computing a diversity score for each of one or more characteristics of the candidate content item, wherein the diversity score is a function of the number of content items in the subset of already presented content items selected to be presented again to the user that share the characteristic with the candidate content item;

based on the diversity scores computed for each content item in the candidate set of content items, selecting one of the content items in the candidate set of content items to present to the user that, relative to the other content items in the candidate set of content items, maximizes entropy with the subset of already presented content items selected to be presented again to the user; and

presenting, to the user, the selected content item and the subset of already presented content items.

2. The method of claim 1, further comprising classifying each of the received content items into one or more groups of content items based at least on a type of the content item.

3. The method of claim 1, further comprising selecting the candidate set of the plurality of content items from the groups of classified content items according to user specified preferences.

4. The method of claim 1, further comprising selecting the candidate set of the plurality of content items from the groups of classified content items based at least on data representing number of past selections of candidate content items from the groups of classified content items by source.

5. The method of claim 1, further comprising:

for each content item in the candidate set of content items, computing a diversity score for each of a plurality of characteristics of the candidate content item, wherein the diversity score is a function of the number of content items in the subset of already presented content items selected to be presented again to the user that share the characteristic with the candidate content item;

for each content item in the candidate set of content items, computing an overall diversity score based on a combination of the diversity scores computed for the content item;

based on the overall diversity scores computed for each content item in the candidate set of content items, selecting one of the content items in the candidate set of content items to present to the user that, relative to the other content items in the candidate set of content items, maximizes entropy with the subset of already presented content items selected to be presented again to the user.

6. A method in a computer system for selecting a content item to present to a user, the method comprising:

obtaining a set of a plurality of content items already presented to the user;

removing at least one content item from the set of content items already presented to the user to select a subset of one or more already presented content items to be presented again to the user;

receiving a plurality of content items not yet presented to the user, each being one of a plurality of different types;

classifying each of the received content items into one or more of a plurality of groups of content items based at least on the type of the content item;

selecting a candidate set of a plurality of content items from the groups of classified content items;

for each content item in the candidate set of content items, computing a diversity score for each of one or more characteristics of the candidate content item, wherein the diversity score is a function of the number of content items in the subset of already presented content items selected to be presented again to the user that share the characteristic with the candidate content item;

based on the diversity scores computed for each content item in the candidate set of content items, selecting one of the content items in the candidate set of content items to present to the user that, relative to the other content items in the candidate set of content items, maximizes entropy with the subset of already presented content items selected to be presented again to the user; and

presenting, to the user, the selected content item and the subset of already presented content items.

7. The method of claim 6, further comprising classifying each of the received content items into one or more groups of content items based at least on a source of the content item.

8. The method of claim 6, further comprising selecting the candidate set of the plurality of content items from the groups of classified content items according to user specified preferences.

9. The method of claim 6, further comprising selecting the candidate set of the plurality of content items from the groups of classified content items based at least on data representing number of past selections of candidate content items from the groups of classified content items by type.

10. The media of claim 6, the method further comprising the steps of:

for each content item in the candidate set of content items, computing a diversity score for each of a plurality of characteristics of the candidate content item, wherein the diversity score is a function of the number of content items in the subset of already presented content items selected to be presented again to the user that share the characteristic with the candidate content item;

for each content item in the candidate set of content items, computing an overall diversity score based on a combination of the diversity scores computed for the content item;

based on the overall diversity scores computed for each content item in the candidate set of content items, selecting one of the content items in the candidate set of content items to present to the user that, relative to the other content items in the candidate set of content items, maximizes entropy with the subset of already presented content items selected to be presented again to the user.

11. One or more non-transitory computer-readable media storing instructions which, when executed by one or more computing devices, cause performance by the one or more computing devices of a method for selecting a content item to present to a user, the method comprising the steps of:

obtaining a set of a plurality of content items already presented to the user;

removing at least one content item from the set of content items already presented to the user to select a subset of one or more already presented content items to be presented again to the user;

receiving a plurality of content items not yet presented to the user, each generated at one of a plurality of different sources;

classifying each of the received content items into one or more of a plurality of groups of content items based at least on the source of the plurality of sources at which the content item is generated;

selecting a candidate set of a plurality of content items from the groups of classified content items;

for each content item in the candidate set of content items, computing a diversity score for each of one or more characteristics of the candidate content item, wherein the diversity score is a function of the number of content items in the subset of already presented content items selected to be presented again to the user that share the characteristic with the candidate content item;

based on the diversity scores computed for each content item in the candidate set of content items, selecting one of the content items in the candidate set of content items to present to the user that, relative to the other content items in the candidate set of content items, maximizes entropy with the subset of already presented content items selected to be presented again to the user; and

presenting, to the user, the selected content item and the subset of already presented content items.

12. The media of claim 11, the method further comprising classifying each of the received content items into one or more groups of content items based at least on a type of the content item.

13. The media of claim 11, the method further comprising selecting the candidate set of the plurality of content items from the groups of classified content items according to user specified preferences.

14. The media of claim 11, the method further comprising selecting the candidate set of the plurality of content items from the groups of classified content items based at least on data representing number of past selections of candidate content items from the groups of classified content items by source.

15. The method of claim 11, further comprising:

for each content item in the candidate set of content items, computing a diversity score for each of a plurality of characteristics of the candidate content item, wherein the diversity score is a function of the number of content items in the subset of already presented content items selected to be presented again to the user that share the characteristic with the candidate content item;

for each content item in the candidate set of content items, computing an overall diversity score based on a combination of the diversity scores computed for the content item;

based on the overall diversity scores computed for each content item in the candidate set of content items, selecting one of the content items in the candidate set of content items to present to the user that, relative to the other content items in the candidate set of content items, maximizes entropy with the subset of already presented content items selected to be presented again to the user.

16. One or more non-transitory computer-readable media storing instructions which, when executed by one or more computing device, cause performance by the one or more computing devices of a method for selecting a content item to present to a user, the method comprising the steps of:

obtaining a set of a plurality of content items already presented to the user;

removing at least one content item from the set of content items already presented to the user to select a subset of one or more already presented content items to be presented again to the user;

receiving a plurality of content items not yet presented to the user, each being one of a plurality of different types;

classifying each of the received content items into one or more of a plurality of groups of content items based at least on the type of the content item;

selecting a candidate set of a plurality of content items from the groups of classified content items;

for each content item in the candidate set of content items, computing a diversity score for each of one or more characteristics of the candidate content item, wherein the diversity score is a function of the number of content items in the subset of already presented content items selected to be presented again to the user that share the characteristic with the candidate content item;

based on the diversity scores computed for each content item in the candidate set of content items, selecting one of the content items in the candidate set of content items to present to the user that, relative to the other content items in the candidate set of content items, maximizes entropy with the subset of already presented content items selected to be presented again to the user; and

presenting, to the user, the selected content item and the subset of already presented content items.

17. The media of claim 16, the method further comprising classifying each of the received content items into one or more groups of content items based at least on a source of the content item.

18. The media of claim 16, the method further comprising selecting the candidate set of the plurality of content items from the groups of classified content items according to user specified preferences.

19. The media of claim 16, the method further comprising selecting the candidate set of the plurality of content items from the groups of classified content items based at least on data representing number of past selections of candidate content items from the groups of classified content items by type.

20. The media of claim 16, the method further comprising the steps of:

for each content item in the candidate set of content items, computing a diversity score for each of a plurality of characteristics of the candidate content item, wherein the diversity score is a function of the number of content items in the subset of already presented content items selected to be presented again to the user that share the characteristic with the candidate content item;

for each content item in the candidate set of content items, computing an overall diversity score based on a combination of the diversity scores computed for the content item;

based on the overall diversity scores computed for each content item in the candidate set of content items, selecting one of the content items in the candidate set of content items to present to the user that, relative to the other content items in the candidate set of content items, maximizes entropy with the subset of already presented content items selected to be presented again to the user.