NLP-based fraud-related short message intelligent identification method and device

By using NLP-based intelligent recognition methods, a text analysis and risk assessment model was constructed, which solved the semantic structure and risk assessment problems in the identification of fraudulent text messages. This enabled accurate understanding and efficient identification of text message content, improving the accuracy of identification and the effectiveness of prevention and control.

CN120994836BActive Publication Date: 2026-06-26GUANGDONG KAITONG SOFTWARE DEV
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
GUANGDONG KAITONG SOFTWARE DEV
Filing Date
2025-09-04
Publication Date
2026-06-26

AI Technical Summary

Technical Problem

Existing methods for identifying fraudulent text messages have shortcomings in text analysis, sequence feature extraction, and risk assessment. They are unable to effectively extract and analyze the semantic structure and contextual features of text messages, and lack the ability to deeply integrate the analysis of the association between phone numbers and website addresses, which affects the accuracy of identification and the effectiveness of prevention and control.

Method used

Employing an NLP-based intelligent recognition method, this approach utilizes text vectorization models, syntactic dependency analysis, scene classification, bidirectional recurrent neural networks, and conditional random field models. By combining residual neural networks and multilayer perceptrons, it achieves accurate understanding and risk assessment of SMS content, and performs correlation analysis and risk classification between phone numbers and website addresses.

Benefits of technology

It significantly improves the accuracy and reliability of identifying fraudulent text messages, enabling precise identification and timely handling of high-risk messages, adapting to changes in fraud methods, and providing continuously optimized protection capabilities.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120994836B_ABST
    Figure CN120994836B_ABST
Patent Text Reader

Abstract

The embodiment of the application provides a kind of based on NLP's method and device for intelligent identification of fraud-related short message, through innovatively constructing text analysis mechanism, through semantic structure extraction and scene classification, realize the accurate understanding of short message content.Design feature extraction model based on sequence labeling, combined with bidirectional recurrent neural network and conditional random field, establish number website association analysis strategy for intelligent identification.Introduce deep fusion evaluation mechanism, through residual neural network and nonlinear mapping, realize the accurate division and timely disposal of risk level.The method effectively solves the deficiencies of traditional technology in text analysis, feature extraction and risk assessment, significantly improves the accuracy and reliability of fraud-related short message identification.
Need to check novelty before this filing date? Find Prior Art