An operating behavior recognition method and device

By converting textual descriptions of merchant operations into image information and using an abnormal behavior recognition model for self-supervised recognition, the problem of low resource consumption and computational efficiency caused by preset strategies in existing technologies is solved, achieving efficient and accurate recognition of risky operational behaviors.

CN115439928BActive Publication Date: 2026-06-26CHINA UNIONPAY
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
CHINA UNIONPAY
Filing Date
2022-08-12
Publication Date
2026-06-26

AI Technical Summary

Technical Problem

Existing technologies require pre-setting multiple risk-related strategies when determining whether a merchant's actions are risky, resulting in high storage resource consumption and slow computation, especially when dealing with large amounts of data.

Method used

The textual descriptions of merchants' operational behaviors are converted into image information and then identified using a trained abnormal behavior recognition model. This model performs self-supervised recognition of operational behaviors, including feature extraction and linear decision-making, avoiding pre-set risky operational behavior strategies and improving recognition efficiency.

Benefits of technology

It improves the accuracy and efficiency of operational behavior recognition, reduces storage resource consumption, adapts to rapid recognition in big data environments, and makes full use of merchants' historical information for longitudinal analysis.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115439928B_ABST
    Figure CN115439928B_ABST
Patent Text Reader

Abstract

Embodiments of the present application provide a kind of operation behavior identification method and device, applied in artificial intelligence technical field, comprising: obtaining the text description information of the current operation behavior of target object, and text description information is converted into to-be-identified image;Through the abnormal behavior identification model that has been trained, to-be-identified image is identified, obtains the preliminary risk score of current operation behavior in multiple dimensions;Based on the preliminary risk score of multiple dimensions, the risk decision result of current operation behavior is obtained.The text description information of the operation behavior of merchant is converted into image information, so that data is more intuitive and more close to monitoring scene, it is convenient for subsequent self-supervision identification to operation behavior using abnormal behavior identification model, determine whether operation behavior is risk behavior, improve the accuracy of operation behavior identification.Furthermore, risk operation behavior strategy is not needed to be preset to identify the operation behavior of merchant, avoid storage resource waste, improve the efficiency of operation behavior identification.
Need to check novelty before this filing date? Find Prior Art