Information sending method, apparatus, device, medium and program product

By constructing a dynamic decision-making mechanism and fusion model, combining reinforcement learning to adjust thresholds, merging small transaction information and embedding marketing content, the problem of high operating costs and lack of commercial value in bank balance change SMS notification services has been solved, achieving resource conservation and improved user experience.

CN122134441APending Publication Date: 2026-06-02INDUSTRIAL AND COMMERCIAL BANK OF CHINA
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
INDUSTRIAL AND COMMERCIAL BANK OF CHINA
Filing Date
2025-08-12
Publication Date
2026-06-02

AI Technical Summary

Technical Problem

The high frequency of SMS notifications for bank balance changes has led to increased operating costs. Existing technology lacks refined control, resulting in redundant high-frequency notifications and wasted resources, while the SMS content lacks commercial value.

Method used

By acquiring users' historical transaction data, a dynamic decision-making mechanism is constructed. A fusion model is used to predict users' transaction behavior, generate an aggregation threshold, and combine it with a reinforcement learning model to adjust the threshold. Small transaction information is merged and marketing information is embedded to improve the conversion rate of SMS marketing.

Benefits of technology

It enabled refined management and control, reduced the number of messages sent, saved computer resources, and improved user experience and SMS marketing conversion rates.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122134441A_ABST
    Figure CN122134441A_ABST
Patent Text Reader

Abstract

This disclosure provides an information sending method. It can be applied to the fields of big data technology and artificial intelligence technology. The method includes: acquiring historical transaction data of a user within a first preset period; extracting features from the historical transaction data to generate user transaction feature data; acquiring the user's current transaction data; predicting the user's transaction behavior based on the current transaction data and the user transaction feature data using a fusion model to generate a first probability value; adjusting an aggregation threshold based on the current transaction data using a reinforcement learning model to generate a target aggregation threshold; generating transaction information based on the first probability value and the target aggregation threshold; inputting the transaction information into a pre-trained product recommendation model to output product recommendation information; and merging the transaction information and product recommendation information into target information and sending it to the user. This disclosure also provides an information sending device, equipment, storage medium, and program product.
Need to check novelty before this filing date? Find Prior Art