Method and system for determining a relevancy parameter for content item

• US 10,674,215 B2
• Filed: 03/29/2019
• Issued: 06/02/2020
• Est. Priority Date: 09/14/2018
• Status: Active Grant

• Current Assignee: Direct Cursus Technology LLC
• Original Assignee: Yandex Europe AG (Yandex N.V.)
• Inventors: Sokolov, Yevgeny Andreevich, Lamburt, Victor Grigorievich, Sharchilev, Boris Dmitrievich, Danilchenko, Andrey Petrovich
• Primary Examiner(s): Huerta, Alexander Q
• Application Number:
  US16/370,286
• Publication Number:
  US 20200092611A1
• Time in Patent Office: 431 Days
• Field of Search: None
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 3/08 Learning methods
• CPC Class Codes: G06N 5/01 Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
• CPC Class Codes: G06N 5/04 Inference or reasoning models
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks
• CPC Class Codes: H04N 21/252 Processing of multiple end-users' preferences to derive collaborative data
• CPC Class Codes: H04N 21/44226 on social networks
• CPC Class Codes: H04N 21/4668 for recommending content, e.g. movies
• CPC Class Codes: H04N 21/4756 for rating content, e.g. scoring a recommended movie
• CPC Class Codes: H04N 21/6582 Data stored in the client, e.g. viewing habits, hardware capabilities, credit card number arrangements where receivers interact with the broadcast H04H20/38

Abstract

A method of determining a relevancy parameter for a digital content item and a system for implementing the method. The digital content item is originated from a content channel associated with a recommendation system. The method is executable by the server. The method comprises: identifying a pool of users associated with the content channel, a given user of the pool of users being associated with the content channel. The method comprises using the pool of users to explore and predict a relevancy parameter. The relevancy parameter is then used for predicting relevancy parameter of the digital content item for a user outside of the pool of users based on the user interactions of the first user.

Claims

1. A method of determining a relevancy parameter for a digital content item, the digital content item being originated from a content channel associated with a recommendation system,

the relevancy parameter for ranking the digital content item as a recommended content item for users of the recommendation system, the recommendation system including a server and at least one client device connectable to the server via a communication network,

the method executable by the server, the server further being configured to execute a recommendation algorithm to generate a set of recommended content items for a given user of the recommendation system;

the method comprising;

identifying a pool of users associated with the content channel, a given user of the pool of users being associated with the content channel;

in response to receiving a content recommendation request from a first client device associated with a first user that belongs to the pool of users;

generating, using the recommendation algorithm, the set of recommended content items for the first user, a given item of the set of recommended content items not originating from the content channel;

artificially inserting into the set of recommended items the digital content item;

gathering an indication of user interactions of the first user with the set of recommended items, the user interactions indicative of the first user'"'"'s propensity for the digital content item; and

predicting relevancy parameter of the digital content item for a user outside of the pool of users based on the user interactions of the first user.

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

in response to receiving a content recommendation requests from other client devices associated with other users that belong to the pool of users;

generating, using the recommendation algorithm, a respective set of recommended items for the other users, a given item of the respective set of recommended items not originating from the content channel;

artificially inserting into the respective set of recommended items the digital content item.

3. The method of claim 2, further comprising observing user interactions of the other users with the respective sets of content recommendations and generating an augmented relevancy parameter associated with the digital content item based on the user interactions of the other users with the respective sets of content recommendations.

4. The method of claim 3, wherein the predicting relevancy parameter of the digital content item for the user outside of the pool of users comprises:

predicting the relevancy parameter based at least in part on the augmented relevancy parameter.

5. The method of claim 4, wherein the augmented relevancy parameter is upwardly biased relative to a native relevancy parameter that would be generated by the recommendation algorithm.

6. The method of claim 1, wherein the method further comprises:

in response to receiving a content recommendation request from a second client device associated with a second user that is outside the pool of users;

generating, using the recommendation algorithm, the set of recommended items for the second user, the set of recommended items including the digital content item, inclusion of the digital content item into the set of recommended items being based on the relevancy parameter;

gathering an indication of user interactions of the second user with the set of recommended items, the user interactions indicative of the second user'"'"'s propensity for the digital content item.

7. The method of claim 6, wherein in response to the user interactions of the second user being indicative of lower propensity for the digital content item of the second user when compared to the first user:

adjusting the relevancy parameter of the digital content item to a lower value thereof.

8. The method of claim 1, wherein the artificially inserting the digital content item into the set of recommended items comprises:

ranking the digital content item relative to other content items within the set of recommended items, the ranking being based on the predicted relevancy parameter of the digital content item.

9. The method of claim 1, wherein the artificially inserting the digital content item into the set of recommended items comprises:

positioning the digital content item relative to other content items within the set of recommended items at a pre-determined position within the set of recommended items.

10. The method of claim 9, wherein the pre-determined position is selected such that to maximize a probability of the user interaction with the digital content item.

11. The method of claim 1, wherein the identifying the pool of users associated with the content channel comprises identifying the given user of the pool of users as being associated with the content channel based on at least one of:

(i) an implicit association;

and (ii) an explicit association.

12. The method of claim 11, wherein the implicit association comprises the first user having been presented with a prior content item from the content channel and the first user not providing an indication of negative propensity in response thereto.

13. The method of claim 11, wherein the explicit association comprises at least one of:

the first user subscribing to the content channel, the user liking a prior content item from the content channel, and the user commenting on the prior content item from the content channel.

14. The method of claim 13, wherein the given item of the set of recommended items is one of a news article, an image, a video, and an interactive snippet.

15. The method of claim 1, wherein the given item of the set of recommended items is originating from a network resource accessible via the communication network.

16. The method of claim 1, wherein all items of the set of recommended items are not originating from the content channel.

17. The method of claim 1, another given item of the set of recommended items is originating from a content channel being one of the content channel and another content channel.

18. The method of claim 1, wherein the content channel is a native channel to the recommendation system.

19. A recommendation server for generating a digital content recommendation, the digital content recommendation for displaying on an electronic device associated with a user, the server connectable to the electronic device via a communication network, the recommendation server executing a ranking algorithm, the recommendation server comprising a processor configured to determine a relevancy parameter for a content item, the digital content item being originated from a content channel associated with the recommendation server,

the relevancy parameter for ranking the digital content item as a recommended content item for users of the recommendation server, the recommendation server including a server and at least one client device connectable to the server via a communication network,

the method executable by the recommendation server, the processor of the recommendation server further being configured to execute a recommendation algorithm to generate a set of recommended content items for a given user of the recommendation server;

the processor being further configured to;

identify a pool of users associated with the content channel, a given user of the pool of users being associated with the content channel;

in response to receiving a content recommendation request from a first client device associated with a first user that belongs to the pool of users;

generate, using the recommendation algorithm, the set of recommended content items for the first user, a given item of the set of recommended content items not originating from the content channel;

artificially insert into the set of recommended items the digital content item;

gather an indication of user interactions of the first user with the set of recommended items, the user interactions indicative of the first user'"'"'s propensity for the digital content item; and

predict a relevancy parameter of the digital content item for a user outside of the pool of users based on the user interactions of the first user.