轻量级AI驱动面向边缘数据流通场景的隐私–效用–效率动态协同优化研究

Research on Lightweight AI-Driven Dynamic Collaborative Optimization of Privacy-Utility-Efficiency in Edge Data Circulation Scenarios

摘要: 随着数字化时代的飞速发展，数据要素作为新质生产力，在数据处理和流通中扮演着日益重要的角色。然而，数据流通面临诸多挑战，如个人数据隐私泄露风险、数据安全效用难以落地以及数据流动审批决策效率低下等问题。在此背景下，中国联合网络通信有限公司软件研究院提出了一种轻量级AI驱动的优化框架，旨在实现隐私、效用与效率的动态协同优化。该框架通过采用轻量级AI数据分类分级技术，对院内各系统、各省分公司的边缘数据进行智能处理，并结合隐私保护、效用评估及效率调控等模块，有效平衡三者关系。实验结果表明，该框架在保障数据隐私的同时，显著提升了数据效用与流通效率，为边缘数据流通领域的发展提供了新的解决方案。

Abstract: With the rapid development of the digital era, data elements, as a new productive force, are playing an increasingly critical role in data processing and circulation. However, data circulation faces multiple challenges, including risks of personal data privacy leakage, difficulties in implementing data security utility, and inefficiencies in data flow approval and decision-making. In this context, the ChinaUnicom Software Research Institute proposes a lightweight AI-driven optimization framework aimed at achieving dynamic collaborative optimization of privacy, utility, and efficiency. This framework employs lightweight AI classification and grading technology to intelligently process edge data from various systems and provincial branches, integrating modules such as privacy protection, utility evaluation, and efficiency regulation to effectively balance these three dimensions. Experimental results demonstrate that the framework ensures data privacy while significantly enhancing data utility and circulation efficiency, providing a novel solution for advancing edge data circulation.