Hebbian learning-based recommendations for social networks

• US 10,362,137 B2
• Filed: 12/28/2015
• Issued: 07/23/2019
• Est. Priority Date: 12/28/2015
• Status: Active Grant

• Current Assignee: Verizon Patent and Licensing Incorporated (Verizon Communications Inc.)
• Original Assignee: Verizon Patent and Licensing Incorporated (Verizon Communications Inc.)
• Inventors: Srinivasaraghavan, Haripriya
• Primary Examiner(s): Hussain, Tauqir
• Assistant Examiner(s): Grijalva Lobos, Boris D
• Application Number:
  US14/979,738
• Publication Number:
  US 20170188101A1
• Time in Patent Office: 1,303 Days
• CPC Class Codes: G06F 11/34 Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
• CPC Class Codes: G06F 11/3438 monitoring of user actions tracking the activity of the user H04L67/535
• CPC Class Codes: H04L 43/04 Processing captured monitoring data, e.g. for logfile generation
• CPC Class Codes: H04L 67/10 in which an application is distributed across nodes in the network software deployment G06F8/60; multiprogramming arrangements G06F9/46
• CPC Class Codes: H04L 67/30 Profiles
• CPC Class Codes: H04L 67/306 User profiles
• CPC Class Codes: H04L 67/53 using third party service providers
• CPC Class Codes: H04N 21/252 Processing of multiple end-users' preferences to derive collaborative data
• CPC Class Codes: H04N 21/25891 being end-user preferences retrieval of video data in a video database based on user preferences G06F16/739; arrangements for recognizing users' preferences H04H60/46; user profiles in network data switching protocols H04L67/306; processing of user preferences or user profiles in wireless networks H04W8/18
• CPC Class Codes: H04N 21/4532 involving end-user characteristics, e.g. viewer profile, preferences monitoring of user activities for profile generation for accessing a video database G06F16/739; user profiles in network data switching protocols H04L67/306; processing of user preferences or user profiles in wireless networks H04W8/18
• CPC Class Codes: H04N 21/4662 characterized by learning algorithms
• CPC Class Codes: H04N 21/4668 for recommending content, e.g. movies
• CPC Class Codes: H04N 21/4788 communicating with other users, e.g. chatting arrangements for providing for computer conferences, e.g. chat rooms, to substation in data switching networks H04L12/1813; distributed application using peer-to-peer [P2P] networks H04L67/104
• CPC Class Codes: H04N 21/4826 using recommendation lists, e.g. of programs or channels sorted out according to their score
• CPC Class Codes: H04N 21/84 Generation or processing of descriptive data, e.g. content descriptors systems specially adapted for using meta-information in broadcast systems H04H60/73

Abstract

A network device applies Hebbian-based learning to provide content recommendations in content-based social networks. The method includes obtaining customer activity data for a content-based social network; modeling the customer activity data as nodes and edges within the content-based social network, the nodes representing users and the edges representing connections between the users; assigning initial weights to the edges, that correspond to a connection strength, based on user-designated of relationships between the nodes; adjusting the initial weights in response to temporally correlated activity between the nodes from the customer activity data, to provide adjusted weights; identifying a content recommendation for a particular node based on an activity to access content by another node and one or more of the adjusted weights; storing a customer profile including the content recommendations associated with a node; and providing the content recommendation to a user device associated with the customer profile.

Claims

1. A method comprising:

obtaining, by a network device, customer activity data for a content-based social network defined by users of a mobile video programming service;

modeling, by the network device, the customer activity data as nodes and edges within the content-based social network, the nodes representing users and the edges representing unidirectional connections between the users, wherein the unidirectional connections between the users are assigned from;

bidirectional relationships based on user-designated social media connections agreed upon by two of the users,

unidirectional relationships based on user-designated social media connections of a single user to receive information from another of the users, and

non-designated relationships between the users;

assigning, by the network device, first initial weights to a first group of the edges based on the bidirectional relationships and the unidirectional relationships;

assigning, by the network device, second initial weights to a second group of the edges based on the non-designated relationships, wherein the second initial weights are smaller than the first initial weights;

adjusting, by the network device, the first and second initial weights in response to temporally correlated selection of a content item from the mobile video programming service by two or more of the nodes, as indicated in the customer activity data, to provide adjusted weights, wherein the adjusting further comprises;

increasing one or more of the first and second initial weights by an amount that is proportional to a difference between an expected outcome of the temporally correlated selection of a content item and an actual outcome of the temporally correlated selection of a content item, or

decreasing one or more of the first and second initial weights by an amount that is proportional to the difference between the expected outcome of the temporally correlated selection of a content item and the actual outcome of the temporally correlated selection of a content item, and wherein the expected outcome is one of;

selecting particular content by the particular node, which was previously selected by the other node, within a correlation time window; and

failing to select the particular content by the particular node, which was previously selected by the other node, within the correlation time window;

identifying, by the network device, a content recommendation for a particular node of the two or more nodes based on an activity to access content by another node of the two or more nodes and one or more of the adjusted weights;

storing, by the network device, a customer profile including the content recommendations associated with the particular node; and

providing, by the network device, the content recommendation to a user device associated with the customer profile.

2. The method of claim 1, wherein the customer activity data includes records of selections of content items from a catalog of available content items and selections of recommended content items.

3. The method of claim 2, wherein the content items include one of television content, video-on-demand content, peer-to-peer content, or advertising content.

4. The method of claim 1, wherein assigning the second initial weights further comprises:

assigning nominal second initial weights, relative to the first initial weights, to the second group of the edges based on the non-designated relationships.

5. The method of claim 1, wherein the customer profile includes one or more of:

a group association of the customer based on the adjusted weights,

an influencer designation, or

an influenced designation.

6. The method of claim 5, wherein providing the content recommendation includes one or more of:

identifying a viewership trend for particular content and providing the content recommendation based on the trend; and

sending the content recommendation to all nodes within a same group as a node that previously selected the particular content.

7. The method of claim 1, wherein the customer profile includes a connection strength threshold for notifications, and

wherein providing the content recommendation includes sending the content recommendation to all nodes having adjusted weights of corresponding edges that exceed the connection strength threshold.

8. The method of claim 1, wherein the customer activity data includes records of selections of content items from a catalog of available content items.

9. The method of claim 1, wherein the customer activity data includes records of selections of recommended content items.

10. A device comprising:

one or more memories to store instructions; and

one or more processing systems to execute the instructions that configure the one or more processing systems to;

obtain customer activity data for a content-based social network defined by users of a mobile video programming service;

model the customer activity data as nodes and edges within the content-based social network, the nodes representing users and the edges representing unidirectional connections between the users, wherein the unidirectional connections between the users are assigned from;

bidirectional relationships based on user-designated social media connections agreed upon by two of the users,

unidirectional relationships based on user-designated social media connections of a single user to receive information from another of the users, and

non-designated relationships between the users;

assign first initial weights to a first group of the edges based on the bidirectional relationships and the unidirectional relationships;

assign second initial weights to a second group of the edges based on the non-designated relationships, wherein the second initial weights are nominal, relative to the first initial weights, based on the non-designated relationships;

adjust the first and second initial weights in response to temporally correlated selection of a content item from the mobile video programming service by two or more of the nodes, as indicated in the customer activity data, to provide adjusted weights, wherein the adjusting includes increasing one or more of the first and second initial weights by an amount that is of a first proportion to a difference between an expected outcome of the temporally correlated selection of a content item and an actual outcome of the temporally correlated selection of a content item, when the expected outcome is selecting particular content, which was previously selected by another node, within a correlation time window;

identify a content recommendation for a particular node of the two or more nodes based on an activity to access content by another node of the two or more nodes and one or more of the adjusted weights;

store, a customer profile including the content recommendations associated with the particular node; and

provide the content recommendation to a user device associated with the customer profile.

11. The device of claim 10, wherein the customer activity data includes records of selections of content items from a catalog of available content items and selections of recommended content items.

12. The device of claim 10, wherein the content item includes advertising content.

13. The device of claim 10, wherein when adjusting the first and second initial weights, the one or more processing systems execute the instructions that further configure the one or more processing systems to:

decrease one or more of the first and second initial weights by an amount that is of a second proportion to a difference between another expected outcome of the temporally correlated selection of a content item and the actual outcome of the temporally correlated selection of a content item, when the other expected outcome is failing to select the particular content, which was previously selected by the other node, within the correlation time window.

14. The device of claim 10, wherein the customer profile includes one or more of:

a group association of the customer based on the adjusted weights,

an influencer designation, or

an influenced designation.

15. The device of claim 10, wherein when providing the content recommendation, the one or more processing systems execute the instructions that further configure the one or more processing systems to:

identify a group, based on the adjusted weights, that includes a node that previously selected particular content, and

send the content recommendation to all nodes within the group.

16. The device of claim 10, wherein the customer activity data includes records of selections of recommended content items.

17. The device of claim 10, wherein the customer activity data includes records of selections of recommended content items.

18. A non-transitory computer-readable medium storing instructions for:

obtaining customer activity data for a content-based social network defined by users of a mobile video programming service;

modeling the customer activity data as nodes and edges within the content-based social network, the nodes representing users and the edges representing unidirectional connections between the users, wherein the unidirectional connections between the users are assigned from;

bidirectional relationships based on user-designated social media connections agreed upon by two of the users,

unidirectional relationships based on user-designated social media connections of a single user to receive information from another of the users, and

non-designated relationships between the users;

assigning first initial weights to a first group of the edges based on the bidirectional relationships and the unidirectional relationships;

assigning second initial weights to a second group of the edges based on the non-designated relationships, wherein the second initial weights are nominal, relative to the first initial weights, based on the non-designated relationships;

adjusting the first and second initial weights in response to temporally correlated selection of a content item from the mobile video programming service by two or more of the nodes, as indicated in the customer activity data, to provide adjusted weights, wherein the adjusting includes increasing one or more of the first and second initial weights by an amount that is of a first proportion to a difference between an expected outcome of the temporally correlated selection of a content item and an actual outcome of the temporally correlated selection of a content item, when the expected outcome is selecting particular content, which was previously selected by another node, within a correlation time window;

identifying a content recommendation for a particular node of the two or more nodes based on an activity to access content by another node of the two or more nodes and one or more of the adjusted weights;

storing a customer profile including the content recommendations associated with the particular node; and

providing the content recommendation to a user device associated with the customer profile.

19. The non-transitory computer-readable medium of claim 18, wherein the instructions for adjusting the initial weights further include instructions for:

increasing one or more of the first and second initial weights by an amount that is proportional to a difference between an expected outcome of the temporally correlated selection of a content item and an actual outcome of the temporally correlated selection of a content item, or

decreasing one or more of the first and second initial weights by an amount that is proportional to the difference between the expected outcome of the temporally correlated selection of a content item and the actual outcome of the temporally correlated selection of a content item,

wherein the expected outcome of the temporally correlated selection of a content item includes a user selection of television content or video-on-demand content that corresponds to the content recommendation.

20. The non-transitory computer-readable medium of claim 18, wherein the customer activity data includes records of selections of content items from a catalog of available content items and selections of recommended content items.