Push model-based activity recommendation method, device, device, and storage medium

Abstract

The embodiment of the present invention discloses a push model-based activity recommendation method, device, device, and storage medium, wherein the method includes the following steps: generating different types of push models based on a big data platform, and selecting one of a plurality of different types of push models The group target model, as the push logic of the current push activity, finally pushes the activity data to the target user based on the target model, and the target user is the user who matches the push logic of the target model. By adopting the present invention, the activity recommendation can be performed according to the recommendation logic of different types of push models, the success rate of the push can be improved, and at the same time, the occurrence of server load overload caused by a large number of inactive users concurrently entering the APP can be avoided.

Description

technical field [0001] The present invention relates to the field of Internet technologies, and in particular, to a method, device, device and storage medium for recommending an activity based on a push model. Background technique [0002] In the daily operation of APP, there are often operating activities such as holidays, new pulls, and events that need to be fully pushed to APP users. During the push, some users will be in an inactive state (that is, the APP has not been launched). After actual measurement, in most cases, the proportion of inactive users far exceeds that of active users, sometimes even as high as tens of hundreds of times. The success rate of reaching the recommended activities (that is, the user sees the recommended activities and opens the related APP) is low. [0003] For activities with greater temptation, inactive users may start APP concurrently, and the APP startup process includes a series of server-side interaction logic such as module initializ...

AI Technical Summary

Problems solved by technology

When pushing, some users will be inactive (that is, the APP is not started). After actual measurement, in most cases, the proportion of inactive users is far higher than that of active users, sometimes even as high as dozens or hundreds of times. The success rate of recommended activities (that is, users see recommended activities and open related APPs) is low

[0003] For activities with greater temptation, inactive users may start the APP concurrently, and the APP startup process includes a series of server-side interaction logic such as module initialization and status verification, which is a process that consumes server resources.

When the push notification arrives, in addition to active users, there may be a large number of inactive users who receive the notification and start the APP, which will cause huge concurrent pressure on the server

Based on the uncertainty of user scale, no matter how hardware resources are optimized, the load limit may be reached, resulting in service unavailability

Images