Multimedia content pushing method and apparatus, computer device, and storage medium

Patent metadata

Field	Value
Publication number	US-12549644-B2
Application number	US-202418759269-A
Country	US
Kind code	B2
Filing date	Jun 28, 2024
Priority date	Jun 30, 2023
Publication date	Feb 10, 2026
Grant date	Feb 10, 2026

Abstract

A multimedia content pushing method and an apparatus, a computer device, and a storage medium are provided. The method includes: obtaining historical operation information of alternative users on a target multimedia content in a target historical time period and real-time operation information on the target multimedia content; determining a target sample user from the alternative users based on the historical operation information and the real-time operation information, and training a neural network model based on sample data composed of sample attribute information corresponding to the target sample user and multimedia attribute information of the target multimedia content to obtain a target neural network model, and in response to a target user of the first type triggering a preset push event, using the target neural network model to determine a push strategy, and pushing the target multimedia content based on the push strategy.

First claim

The invention claimed is:

1 . A multimedia content pushing method, comprising:
obtaining historical operation information of a plurality of alternative users on a target multimedia content in a target historical time period and real-time operation information of the alternative users on the target multimedia content; determining a target sample user from the plurality of alternative users based on the historical operation information and the real-time operation information of the plurality of alternative users, and training a neural network model to be trained based on sample data composed of sample attribute information corresponding to the target sample user and multimedia attribute information of the target multimedia content to obtain a target neural network model, wherein the target sample user comprises an alternative user whose user type is converted from a first type to a second type in a case where the target multimedia content is pushed thereto; and in response to a target user of the first type triggering a preset push event, determining a push strategy of pushing the target multimedia content to the target user using the target neural network model, and pushing the target multimedia content based on the push strategy.

2 . The multimedia content pushing method according to claim 1 , wherein the obtaining historical operation information of a plurality of alternative users on a target multimedia content in a target historical time period comprises:
determining object attribute information of a content object corresponding to the target multimedia content; and determining the target historical time period corresponding to the target multimedia content based on the object attribute information, and obtaining the historical operation information of the plurality of alternative users on the target multimedia content in the target historical time period.

3 . The multimedia content pushing method according to claim 1 , wherein obtaining real-time operation information of the alternative users on the target multimedia content comprises:
for each of the plurality of alternative users and in each of a plurality of operation cycles corresponding to the alternative user, determining the operation information of the alternative user on the target multimedia content in each operation cycle; and obtaining the real-time operation information of each of the plurality of alternative users based on the operation information respectively corresponding to the plurality of operation cycles.

4 . The multimedia content pushing method according to claim 1 , wherein the determining a target sample user from the plurality of alternative users based on the historical operation information and the real-time operation information of the plurality of alternative users comprises:
for each of the plurality of alternative users, fusing the historical operation information and the real-time operation information of the alternative user to obtain target operation information of the alternative user; and in response to a target operation information of any alternative user meeting a preset condition, determining the alternative user as the target sample user, wherein the preset condition satisfies that the alternative user is converted from the first type to the second type.

5 . The multimedia content pushing method according to claim 4 , wherein the first type comprises a shallow interaction group; the second type comprises an influenced group; and
the preset condition comprises at least one of follows: an exposure count of the target multimedia content being greater than or equal to a first exposure count threshold, and an exposure count with an exposure duration longer than a first preset duration threshold being greater than a second exposure count threshold; a play duration of the target multimedia content being longer than a first play count threshold, and a play count with a play duration longer than a second preset duration threshold being greater than a preset play count threshold; a count of clicks made by the alternative user on the target multimedia content being greater than or equal to a preset click count threshold; a stay duration of an advertisement landing page corresponding to the target multimedia content being longer than a preset stay duration threshold; a count of likes given by the alternative user for the target multimedia content being greater than a preset like count threshold; a count of comments made by the alternative user on the target multimedia content being greater than or equal to a comment count threshold; and a count of sharing the target multimedia content by the alternative user being greater than or equal to a sharing count threshold.

6 . The multimedia content pushing method according to claim 1 , wherein the in response to a target user of the first type triggering a preset push event, determining a push strategy of pushing the target multimedia content to the target user using the target neural network model comprises:
in response to the target user of the first type triggering the preset push event, using the target neural network model to determine a prediction result about the target user being converted from the first type to the second type after pushing the target multimedia content to the target user; and determining the push strategy of pushing the target multimedia content to the target user based on the prediction result.

7 . The multimedia content pushing method according to claim 6 , wherein the using the target neural network model to determine a prediction result about the target user being converted from the first type to the second type after pushing the target multimedia content to the target user comprises:
obtaining attribute information corresponding to the target user; and constituting data to be processed based on the attribute information and the multimedia attribute information of the target multimedia content, and inputting the data to be processed to the target neural network model to obtain the prediction result.

8 . The multimedia content pushing method according to claim 1 , after pushing the target multimedia content to the target user, further comprising:
obtaining historical operation information and real-time operation information of the target user on the target multimedia content; and determining a conversion result of the target user from the first type to the second type with the historical operation information and the real-time operation information of the target user on the target multimedia content.

9 . The multimedia content pushing method according to claim 8 , further comprising:
using the conversion result to retrain the target neural network model.

10 . The multimedia content pushing method according to claim 1 , wherein the training a neural network model to be trained based on sample data composed of sample attribute information corresponding to the target sample user and multimedia attribute information of the target multimedia content to obtain a target neural network model, comprises:
taking the sample data as positive sample data; performing random sampling on non-target sample users of the plurality of alternative users and constituting negative sample data with sample attribute information corresponding to the randomly sampled alternative users and the multimedia attribute information of the target multimedia content, to maintain a certain ratio of the positive sample data and the negative sample data, and training the neural network to be trained with the positive sample data and the negative sample data to obtain the target neural network model.

11 . A computer device, compri

Assignees

• Beijing Youzhuju Network Tech Co Ltd

Inventors

• He Pan
• Han Feng
• Chen Mutong
• Yu Senyang
• Wang Kunyang

Classifications

• G06Q30/0277

  Online advertisement · CPC title

• G06Q30/0255

  based on user history · CPC title

• G06N3/08

  Learning methods · CPC title

• G06F18/25

  Fusion techniques · CPC title

• H04L67/535 Primary

  Tracking the activity of the user (network monitoring arrangements H04L43/00; recording of computer activity G06F11/34) · CPC title