API request classification method and device, computer equipment and storage medium
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
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.
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.
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.
Smart Images

Figure CN121859244A_ABST