Resource recommendation method and device based on recommendation model, electronic equipment and medium

CN116450944BActive Publication Date: 2026-06-26BEIJING BAIDU NETCOM SCI & TECH CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
BEIJING BAIDU NETCOM SCI & TECH CO LTD
Filing Date
2023-04-18
Publication Date
2026-06-26

AI Technical Summary

Technical Problem

Existing recommendation systems perform poorly in recommending candidate resources in different scenarios, failing to effectively characterize the features of target users, resulting in insufficient recommendation accuracy.

Method used

A recommendation model-based approach is adopted, which utilizes multiple expert networks and multiple gating networks to correspond to multiple user groups and multiple scenarios, respectively. Through feature information extraction and recommendation weight calculation, it is determined whether to recommend candidate resources in a specific scenario.

Benefits of technology

It improves the accuracy of resource recommendations, especially in scenarios involving the first screen and non-first screens. It can independently optimize the strategy framework, thereby improving user experience and the success rate of resource recommendations.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116450944B_ABST
    Figure CN116450944B_ABST
Patent Text Reader

Abstract

The present disclosure provides a resource recommendation method and device based on a recommendation model, electronic equipment and a medium, relating to the technical field of artificial intelligence, in particular to the technical field of deep learning. The implementation scheme is: obtaining behavior data of a user and a candidate resource; extracting feature information based on the behavior data and the candidate resource; inputting the feature information into a plurality of expert networks to obtain a first output of the expert networks, wherein the first output has a feature representation of a corresponding user group; inputting the first output of the expert networks into a plurality of gate networks respectively to obtain a second output of the gate networks, wherein the second output includes a recommendation weight of the candidate resource for a plurality of recommendation factors in a corresponding scene; for at least one scene of a plurality of scenes, determining whether to recommend the candidate resource in the scene based on the recommendation weight of the plurality of recommendation factors.
Need to check novelty before this filing date? Find Prior Art