Intelligent matching method of 5g message rich media content and user interest label

By constructing a three-dimensional temporal signal vector and a lightweight LSTM-CNN hybrid model, and dynamically adjusting the weights of popularity and interest, the problem of insufficient user state perception in 5G message recommendation systems under high-concurrency scenarios is solved, thereby improving the diversity and relevance of content and adapting recommendation strategies to different load states.

CN122240862APending Publication Date: 2026-06-19GUANGDONG BOJIN INFORMATION TECHNOLOGY GROUP CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
GUANGDONG BOJIN INFORMATION TECHNOLOGY GROUP CO LTD
Filing Date
2026-03-23
Publication Date
2026-06-19

AI Technical Summary

Technical Problem

Existing 5G messaging rich media content recommendation systems struggle to perceive user status in real time under high concurrency, light interaction, and time-sensitive scenarios, resulting in ranking results biased towards popular content or single interests, affecting information diversity and causing user cognitive fatigue. Furthermore, the complex algorithms increase the difficulty of system deployment and are not suitable for edge devices.

Method used

By constructing a three-dimensional temporal signal vector, combining a lightweight LSTM-CNN hybrid model and a piecewise nonlinear mapping function, the weights of popularity and interest are dynamically adjusted. By utilizing the cognitive load index and context-sensitive attention mechanism, real-time matching and adaptive ranking of user interests and content popularity are achieved.

Benefits of technology

It improves the relevance and diversity of recommended content, enhances the system's robustness and context awareness in complex 5G messaging scenarios, ensures that the recommendation strategy is adjusted in a timely manner under different load conditions, and improves user experience and system response sensitivity.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122240862A_ABST
    Figure CN122240862A_ABST
Patent Text Reader

Abstract

This invention relates to an intelligent matching method for rich media content in 5G messaging and user interest tags. The invention collects multi-source lightweight user behavior signals, constructs a lightweight LSTM-CNN hybrid model with a three-dimensional time-series vector input trained under eye-tracking experiments and subjective load labeling supervision, and outputs a continuous cognitive load index. Based on this index, it segments and maps popularity and interest weights, calculating the interest matching score and popularity score of the rich media content respectively. After fusion, it compensates for ranking bias and generates the final recommendation priority. When user response and behavior performance are below a preset baseline, the Federated Distillation method is used to fine-tune the model parameters online. This invention achieves adaptive dynamic coordination between user cognitive load and content push weights, improving the relevance of recommended content and user experience.
Need to check novelty before this filing date? Find Prior Art