Abnormal Online Behavior Detection Using Multi-Dimensional Feature Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing abnormal online operational behavior detection methods face challenges in accuracy due to low-dimensional features and high feature dimension complexity, leading to errors in detection results and poor real-time performance.
Innovation Solution
The proposed method involves acquiring online operation behavior information, obtaining sub-models from preset object models, determining abnormal data volumes and maximum data volumes, and comparing these volumes to determine detection results, thereby improving accuracy in abnormal behavior detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If clustering processing is performed based on low-dimensional features to determine detection results, then the device complexity is reduced, but the measurement precision of abnormal behavior detection deteriorates
Solution Approach 1:
The patent transforms the low-dimensional feature space into a high-dimensional feature space by introducing multiple new dimensions including time dimension (historical behavior sequences), spatial dimension (behavior relationships between different objects), and attribute dimension (multiple behavior characteristics). This dimensional expansion enables more accurate detection while maintaining manageable system complexity through structured feature engineering.
2Measurement precision
If high-dimensional features are used to improve detection accuracy, then the measurement precision improves, but the device complexity increases
Solution Approach 1:
The patent segments the complex high-dimensional feature space into multiple independent modules: user behavior features, merchant behavior features, transaction features, device features, and temporal features. Each module processes specific aspects independently, then their results are integrated. This segmentation reduces the computational burden of handling all dimensions simultaneously while preserving the benefits of high-dimensional analysis.
3Measurement precision
If complex feature processing is performed to improve detection accuracy, then the measurement precision improves, but the productivity of real-time detection deteriorates
Solution Approach 1:
The patent performs preliminary processing of behavior features by pre-computing user behavior profiles, merchant behavior profiles, and establishing baseline patterns during normal operation. When anomalies need to be detected, the system only needs to compare current behavior against these pre-established patterns rather than performing full complex analysis, significantly improving real-time detection speed while maintaining accuracy.
Data Source
AI summary
This application provides an abnormal online operational behavior detection method performed by an electronic device. The method includes: obtaining a first target sub-model corresponding to a first target object from a first preset object model; determining an abnormal data volume from the first target sub-model based on a preset model parameter and a first detection result by comparing a target data volume and the abnormal data volume; obtaining a second target sub-model corresponding to a second target object and having a highest similarity with the first target sub-model from a second preset object model; obtaining a target maximum data volume corresponding to the second target sub-model, and determining a second detection result by comparing the target data volume and the target maximum data volume; and determining a target detection result of the online operation behavior information in accordance with the first detection result and the second detection result.