A Multi-Agent Collaborative Data Processing Method and System for E-commerce Operations

By quantifying multi-agent decision-making into decision vectors for traffic, profit, and risk and weighting and fusing them, and adjusting the scoring thresholds based on the characteristics of the operational stage, the problem of decision conflict in multi-agent systems is solved, and efficient and intelligent e-commerce operations are achieved.

CN122312263APending Publication Date: 2026-06-30SICHUAN YUANSHENGHUI DIGITAL TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-05-29
Publication Date
2026-06-30

AI Technical Summary

Technical Problem

Existing multi-agent systems lack automated arbitration mechanisms when decision-making conflicts occur, leading to a disconnect between automated decisions and the goals of senior management, resulting in low-quality decisions.

Method used

By collecting multi-source heterogeneous operational data, reinforcement learning models are used to quantify the decision suggestions of each agent into decision vectors for traffic, profit, and risk. A comprehensive score is calculated through weighted fusion, and the preset score threshold is dynamically adjusted in combination with the characteristics of the operational stage to achieve collaborative decision-making among multiple agents.

Benefits of technology

It provides objective and calculable solutions to decision conflicts, ensuring decision quality, balancing growth and risk control, and enabling intelligent operation in complex e-commerce environments.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122312263A_ABST
    Figure CN122312263A_ABST
Patent Text Reader

Abstract

This invention discloses a multi-agent collaborative data processing method and system for e-commerce operations, relating to the fields of internet data processing and artificial intelligence. The method includes: multiple agents accessing multi-source heterogeneous operational data and outputting decision suggestions; when conflicting decision suggestions exist between agents, each suggestion is quantified into a three-dimensional decision vector containing traffic, profit, and risk; the current operational stage and operational status characteristics of the system are input into a reinforcement learning model to obtain weights for each dimension; based on the obtained weights, each suggestion is weighted and fused to obtain a comprehensive score; if the highest score is greater than a preset score threshold, the corresponding suggestion is executed; if it is less than or equal to the threshold, feedback is provided requesting the agent to re-suggest. The score threshold is dynamically adjusted according to the operational stage, with the threshold for the profitable period being higher than that for the initial launch period. This method solves the problem of decision conflict among multiple agents and achieves an intelligent balance between growth, profitability, and risk control in e-commerce operations.
Need to check novelty before this filing date? Find Prior Art