Predicting content and context performance based on performance history of users

• US 8,812,494 B2
• Filed: 01/04/2013
• Issued: 08/19/2014
• Est. Priority Date: 05/28/2010
• Status: Active Grant

• Current Assignee: Apple Inc.
• Original Assignee: Apple Inc.
• Inventors: Grigorovici, Dan Marius, Chittari, Ravikiran, Mukherjee, Prasenjit, Abdala, Omar, Priyadershan, Eswar
• Primary Examiner(s): Mahmoudi, Tony
• Assistant Examiner(s): NGUYEN, MERILYN P
• Application Number:
  US13/734,723
• Publication Number:
  US 20130132391A1
• Time in Patent Office: 592 Days
• Field of Search: 707/723, 707/737, 707/748, 707/749, 707/751, 707/752, 707/722
• US Class Current: 707/722
• CPC Class Codes: G06F 16/285 Clustering or classification
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0201 Market modelling; Market analysis; Collecting market data
• CPC Class Codes: G06Q 30/0202 Market predictions or forecasting for commercial activities
• CPC Class Codes: G06Q 30/0282 Rating or review of business operators or products

Abstract

Systems and methods are provided for selecting contexts for new invitational content and invitational content for new contexts. In particular, a performance history of delivered invitational content in known contexts is combined with similarity measures for the delivered invitational content, with respect to a new invitational content, to generate a list of potential contexts for the new invitational content. Similarly, a performance history of in known contexts with delivered invitational content can combined with similarity measures for known contexts, with respect to a new context, to generate a list of potential content for the new context. Further, a combination of these methods can be used to pair new invitational content with new contexts.

Claims

1. A method, comprising:

assembling, via a processor, an input data set based on performance data of delivered invitational content with respect to known contexts, content similarity data for the delivered invitational content, and context similarity data for the known contexts;

identifying clusters in the input data set, each of the clusters associating at least one of the delivered invitational content and at least one of the known contexts;

generating first rank values for the clusters with respect to at least one of the delivered invitational content and second rank values for the clusters with respect to at least one of the known contexts;

computing total rank values for the identified known contexts, wherein for each of the identified known contexts the computing comprises;

identifying one or more rank paths associated with the one of the identified known contexts,

calculating a product of first and second rank values associated with each of the one or more rank paths, and

calculating a total rank value for the one of the identified known contexts as a sum of the rank path products for each of the one or more rank paths; and

storing a dataset for a database, the dataset comprising at least the delivered invitational content, the known contexts, the identified clusters, the first rank values, the second rank values, the total rank values, content metadata for the delivered invitational content, and context metadata for the known contexts.

2. The method of claim 1, wherein the step of assembling further comprises:

computing content similarity values for the delivered invitational content based on the content metadata; and

computing context similarity values for known contexts based on the context metadata,

wherein each of the content and context similarity values comprises a Jaccard similarity value.

3. The method of claim 1, wherein the step of identifying further comprises:

determining an initial membership for the clusters from the input data set;

obtaining a probability function based on the initial membership; and

defining the clusters to include the delivered invitational content and the known contexts within a distance based on moments of the obtained probability function.

4. The method of claim 3, wherein the step of generating the first and second rank values comprises:

computing the first and second rank values for each of the clusters based on the moments of the obtained probability function.

5. The method of claim 1, wherein the first rank values measure a relation between the clusters and the at least one of the delivered invitational content, and the second rank values measure a relation between the clusters and the at least one of the known contexts.

6. The method of claim 1, further comprising:

generating an ordered list of identified contexts for a candidate invitational content item based on the clusters, the first rank values, and the second rank values.

7. The method of claim 1, further comprising:

generating an ordered list of delivered content items for a candidate context based on the clusters, the first rank values, and the second rank values.

8. The method of claim 1, wherein the one or more rank paths comprises a one of the identified clusters associated with a one of the identified known contexts and a one of the identified delivered invitational content.

9. A system, comprising:

a processor;

a data importation element configured to control the processor to assemble an input data set based on performance data for delivered invitational content with respect to known contexts, content metadata for the delivered invitational content, and context metadata for the known contexts;

a cluster extraction element configured to control the processor to identify clusters based on the input data, each of the clusters associating at least one of the delivered invitational content and at least one of the known contexts;

a data exporter element configured to control the processor to generate first rank values for the clusters with respect to the at least one of the delivered invitational content, generating second rank values for the clusters with respect to the at least one of the known contexts, computing total rank values for the identified known contexts, and storing a dataset for a database, the dataset comprising at least the delivered invitational content, the known contexts, the identified clusters, the first rank values, the second rank values, the total rank values, content metadata for the delivered invitational content, and context metadata for the known contexts, wherein computing total rank values for each of the identified known contexts comprises;

identifying one or more rank paths associated with the one of the identified known contexts,

calculating a product of first and second rank values associated with each of the one or more rank paths, and

calculating a total rank value for the one of the identified known contexts as a sum of the rank path products for each of the one or more rank paths.

10. The system of claim 9, wherein the data importation element is further configured to control the processor to:

compute content similarity values for the delivered invitational content based on the content metadata; and

compute context similarity values for existing contexts based on the context metadata; and

combine the performance data, the content similarity values, and the context similarity values to form the input data set.

11. The system of claim 9, wherein the cluster extraction element is further configured to control the processor to determine an initial membership for the clusters from the input data set using an extended Fiedler method and defining the clusters to include the delivered invitational content and the known contexts within a distance based on the moments of a Gaussian probability function based on the initial membership.

12. The system of claim 9, wherein the data exporter element is further configured to control the processor to generate the first and second rank values by:

determining statistical characteristics for each of the clusters; and

computing the first and second rank values for each of the clusters based on the statistical characteristics.

13. The system of claim 9, wherein the first rank values measure a relation between the clusters and the at least one of the delivered invitational content, and the second rank values measure a relation between the clusters and the at least one of the known contexts.

14. The system of claim 9, further comprising:

a request processing element configured to generate an ordered list of identified contexts for a candidate invitational content item based on the clusters, the first rank values, and the second rank values.

15. The system of claim 9, further comprising:

a request processing element configured to generate an ordered list of delivered content items for a candidate context based on the clusters, the first rank values, and the second rank values.

16. A non-transitory computer-readable medium having code for causing a computer to perform a method stored thereon, the method comprising:

assembling, via a processor, an input data set based on performance data of delivered invitational content with respect to known contexts, content similarity data for the delivered invitational content, and context similarity data for the known contexts;

identifying clusters in the input data set, each of the clusters associating at least one of the delivered invitational content and at least one of the known contexts;

generating first rank values for the clusters with respect to at least one of the delivered invitational content and second rank values for the clusters with respect to at least one of the known contexts, wherein the first rank values measure a relation between the clusters and the at least one of the delivered invitational content, and the second rank values measure a relation between the clusters and the at least one of the known contexts;

computing total rank values for the identified known contexts, wherein for each of the identified known contexts the computing comprises;

identifying one or more rank paths associated with the one of the identified known contexts,

calculating a product of first and second rank values associated with each of the one or more rank paths, and

calculating a total rank value for the one of the identified known contexts as a sum of the rank path products for each of the one or more rank paths; and

storing a dataset for a database, the dataset comprising at least the delivered invitational content, the known contexts, the identified clusters, the first rank values, the second rank values, the total rank values, content metadata for the delivered invitational content, and context metadata for the known contexts.

17. The non-transitory computer-readable medium of claim 16, wherein the step of assembling further comprises:

computing content similarity values for the delivered invitational content based on the content metadata; and

computing context similarity values for known contexts based on the context metadata,

wherein each of the content and context similarity values comprises a Jaccard similarity value.

18. The non-transitory computer-readable medium of claim 16, wherein the step of identifying further comprises:

determining an initial membership for the clusters from the input data set;

obtaining a probability function based on the initial membership;

defining the clusters to include the delivered invitational content and the known contexts within a distance based on moments of the obtained probability function; and

computing the first and second rank values for each of the clusters based on the moments of the obtained probability function.

19. The non-transitory computer-readable medium of claim 16, further comprising:

generating an ordered list of identified contexts for a candidate invitational content item based on the clusters, the first rank values, and the second rank values.

20. The non-transitory computer-readable medium of claim 16, further comprising:

generating an ordered list of delivered content items for a candidate context based on the clusters, the first rank values, and the second rank values.