API request classification method and device, computer equipment and storage medium

CN121859244APending Publication Date: 2026-04-14SHENZHEN YISHIHUOLALA TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
SHENZHEN YISHIHUOLALA TECH CO LTD
Filing Date
2025-12-30
Publication Date
2026-04-14

AI Technical Summary

Technical Problem

Existing API request classification methods suffer from insufficient feature discrimination, poor scenario adaptability, weak adaptability to data imbalance, and insufficient utilization of request header information. This results in low recall rates for rare classes, classification results biased towards the majority class, and high computational resource requirements, making it difficult to meet the deployment needs of small-scale datasets.

Method used

By obtaining the business requests of the target API, we perform structured parsing, extract structured features, business attribute features, and general attribute features, obtain a keyword list in combination with the business scenario, generate semantic features and perform feature fusion, and use an ensemble tree model classifier for classification processing to generate classification labels and confidence scores.

Benefits of technology

It significantly improves the recall rate and classification accuracy of rare categories, enhances the adaptability to scenarios with heterogeneous API request structures, reduces customized maintenance costs, optimizes the stability and reliability of classification results, improves classification accuracy, and simplifies computing resource requirements.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121859244A_ABST
    Figure CN121859244A_ABST
Patent Text Reader

Abstract

The invention discloses an API request classification method and device, computer equipment and a storage medium, and the method comprises the steps: obtaining a business request corresponding to a target API, carrying out the structural analysis and information extraction of the business request, and obtaining original business information and business key features; obtaining a corresponding service keyword list based on a service scene corresponding to the service request; matching original business information with the business keyword list, and generating semantic features according to a matching result; carrying out feature fusion on the semantic features and the business key features to obtain fusion features; and inputting the fusion feature into a preset classification model for classification processing to obtain a classification tag corresponding to the target API and a corresponding confidence coefficient. In the embodiment of the invention, through dynamic keyword matching and multi-feature fusion, the recall rate of rare categories can be improved, the suitability of heterogeneous request scenes is enhanced, meanwhile, the problem of data imbalance is effectively relieved, and the stability and accuracy of model classification are optimized.
Need to check novelty before this filing date? Find Prior Art