System and method for personalized content recommendations

• US 10,521,824 B1
• Filed: 01/02/2014
• Issued: 12/31/2019
• Est. Priority Date: 01/02/2014
• Status: Active Grant

• Current Assignee: Outbrain Incorporated
• Original Assignee: Outbrain Incorporated
• Inventors: Boshy, Shlomy, Galai, Yaron
• Primary Examiner(s): Schmucker, Michael W
• Assistant Examiner(s): Shelden, Bion A
• Application Number:
  US14/146,279
• Time in Patent Office: 2,189 Days
• Field of Search: None
• CPC Class Codes: G06Q 30/0269 based on user profile or attribute

Abstract

Identifying personalized content recommendations for users in an electronic environment is disclosed. User data comprising information relating to web-based content consumption of multiple users is collected. Multiple user cluster types associated with one or more interest categories are established. A feature vector is generated for each user for each of the multiple user cluster types. Based on the generated feature vectors, the user are grouped into multiple clusters. A grade is generated for each of a plurality of candidate recommendations for each of the clusters. Based on the generated grades, one or more personalized content recommendations for each of the clusters are identified.

Claims

1. A method comprising:

collecting, via a network, user data associated with a plurality of users, wherein the user data comprises information relating to web-based interactions with content by the plurality of users;

assigning a weight to each of the web-based interactions of the user data in view of an age of the web-based interactions;

defining a plurality of user cluster types, wherein each user cluster type is associated with one or more interest categories;

generating, for each of the plurality of users, a vector representation corresponding to each of the plurality of user cluster types, wherein the vector representation comprises a weighted value representing an interest level of a user corresponding to each of the one or more interest categories associated with each user cluster type;

normalizing the vector representation based on an editorial bias representing a level of consumption of content associated with the one or more interest categories as a function of placement of the content on a webpage;

determining a portion of the user data matches a first vector representation associated with a first cluster type based at least in part on the weight assigned to the web-based interactions, the first vector representation comprising a first weighted value for a first interest category associated with the first cluster type;

defining a first cluster comprising a first plurality of users of the plurality of users, wherein the first plurality of users are associated with the first user data;

generating a plurality of grades for a plurality of content recommendations, wherein a first grade of the plurality of grades is based on one or more user engagement indicators associated with the first plurality of users in the first cluster and a first content recommendation of the plurality of content recommendations;

identifying, based on a comparison of the plurality of grades, the first content recommendation to provision to the first cluster, wherein the first grade corresponds to a first engagement level of the first cluster as it relates to the first content recommendation;

loading, into a memory, the first content recommendation in association with the first cluster and a first publisher; and

in response to a target user included in the first cluster accessing a webpage provided by the first publisher;

adjusting the first grade associated with the first content recommendation and the first cluster comprising the first plurality of users based on user property data, user-specific action data and non-action data associated with the target user, wherein the user property data comprises a geographic location of the target user, wherein the action data identifies a plurality of web-based activities executed by the target user, and wherein the non-action data identifies passively generated data comprising information associated with the target user viewing a content link with no interaction with the content link; and

populating, in view of the first grade, a first designated portion of the webpage accessed by the target user with the first content recommendation associated with the first cluster.

2. The method of claim 1, further comprising determining a quantity of visits to the first content recommendation by users in the first cluster, wherein the first grade for the first content recommendation for the first cluster is based on the quantity of visits.

3. The method of claim 1, further comprising identifying the first content recommendation for the target user based on user-specific data associated with the target user identified from the user data.

4. A system comprising:

a memory to store instructions; and

a processing device coupled with the memory, the processing device configured to execute the instructions to;

collect, via a network, user data associated with a plurality of users, wherein the user data comprises information relating to web-based interactions with content by the plurality of users;

assign a weight to each of the web-based interactions of the user data in view of an age of the web-based interactions;

define a plurality of user cluster types, wherein each user cluster type is associated with one or more interest categories;

generate, for each of the plurality of users, a vector representation corresponding to each of the plurality of user cluster types, wherein the vector representation comprises a weighted value representing an interest level of a user corresponding to each of the one or more interest categories associated with each user cluster type;

normalize the vector representation based on an editorial bias representing a level of consumption of content associated with the one or more interest categories as a function of placement of the content on a webpage;

determine a portion of the user data matches a first vector representation associated with a first cluster type based at least in part on the weight assigned to the web-based interactions, the first vector representation comprising a first weighted value for a first interest category associated with the first cluster type;

define a first cluster comprising a first plurality of users of the plurality of users, wherein the first plurality of users are associated with the first user data;

generate a plurality of grades for a plurality of content recommendations, wherein a first grade of the plurality of grades is based on one or more user engagement indicators associated with the first plurality of users in the first cluster and a first content recommendation of the plurality of content recommendations;

identify, based on a comparison of the plurality of grades, the first content recommendation to provision to the first cluster, wherein the first grade corresponds to a first engagement level of the first cluster as it relates to the first content recommendation;

load, into a memory, the first content recommendation in association with the first cluster and a first publisher; and

in response to a target user included in the first cluster accessing a webpage provided by the first publisher;

adjust the first grade associated with the first content recommendation and the first cluster comprising the first plurality of users based on user property data, user-specific action data and non-action data associated with the target user, wherein the user property data comprises a geographic location of the target user, wherein the action data identifies a plurality of web-based activities executed by the target user, and wherein the non-action data identifies passively generated data comprising information associated with the target user viewing a content link with no interaction with the content link; and

populate, in view of the first grade, a first designated portion of the webpage accessed by the target user with the first content recommendation associated with the first cluster.

5. The system of claim 4, the processing device further configured to:

determine a quantity of visits to the first content recommendation by users in the first cluster, wherein the first grade for the first content recommendation for the first cluster is based on the quantity of visits.

6. The system of claim 4, the processing device further configured to:

identify the first content recommendation for the target user based on user-specific data associated with the target user identified from the user data.

7. A non-transitory computer readable storage medium comprising instructions thereon that, in response to execution by a processing device, cause the processing device to perform operations comprising:

collecting, via a network, user data associated with a plurality of users, wherein the user data comprises information relating to web-based interactions with content by the plurality of users;

assigning a weight to each of the web-based interactions of the user data in view of an age of the web-based interactions;

defining a plurality of user cluster types, wherein each user cluster type is associated with one or more interest categories;

generating, for each of the plurality of users, a vector representation corresponding to each of the plurality of user cluster types, wherein the vector representation comprises a weighted value representing an interest level of a user corresponding to each of the one or more interest categories associated with each user cluster type;

normalizing the vector representation based on an editorial bias representing a level of consumption of content associated with the one or more interest categories as a function of placement page;

determining a portion of the user data matches a first vector representation associated with a first cluster type based at least in part on the weight assigned to the web-based interactions, the first vector representation comprising a first weighted value for a first interest category associated with the first cluster type;

defining a first cluster comprising a first plurality of users of the plurality of users, wherein the first plurality of users are associated with the first user data;

generating a plurality of grades for a plurality of content recommendations, wherein a first grade of the plurality of grades is based on one or more user engagement indicators associated with the first plurality of users in the first cluster and a first content recommendation of the plurality of content recommendations;

identifying, based on a comparison of the plurality of grades, the first content recommendation to provision to the first cluster, wherein the first grade corresponds to a first engagement level of the first cluster as it relates to the first content recommendation;

loading, into a memory, the first content recommendation in association with the first cluster and a first publisher; and

in response to a target user included in the first cluster accessing a webpage provided by the first publisher;

adjusting the first grade associated with the first content recommendation and the first cluster comprising the first plurality of users based on user property data, user-specific action data and non-action data associated with the target user, wherein the user property data comprises a geographic location of the target user, wherein the action data identifies a plurality of web-based activities executed by the target user, and wherein the non-action data identifies passively generated data comprising information associated with the target user viewing a content link with no interaction with the content link; and

populating, in view of the first grade, a first designated portion of the webpage accessed by the target user with the first content recommendation associated with the first cluster.

8. The non-transitory computer readable storage medium of claim 7, the operations further comprising:

determining a quantity of visits to the first content recommendation by users in the first cluster, wherein the first grade for the first content recommendation for the first cluster is based on the quantity of visits.

9. The non-transitory computer readable storage medium of claim 7, the operations further comprising:

identifying the first content recommendation for the target user based on user-specific data associated with the target user identified from the user data.