利用机器学习技术防范网络意识形态风险的理论模型与逻辑进路  

Theoretical Model and Logical Approach for Mitigating Online Ideological Risks through Machine Learning Techniques

作　　者：秦博 徐浩铭[3] QIN Bo;XU Hao-Ming

机构地区：[1]电子科技大学外国语学院,四川成都611731 [2]电子科技大学东南亚研究所,四川成都611731 [3]中共成都市委党校科技与生态文明教研部,四川成都610110

出　　处：《党政研究》2024年第4期4-13,124,共11页Studies on Party and Government

基　　金：国家社科基金项目“中国科技形象在推特的计算传播对策研究”(20CXW016)。

摘　　要：新媒介时代网络日益发达,对我国意识形态的安全和治理带来了巨大挑战。现代机器学习技术特别是那些基于深度神经网络的模型,能够从大量非结构化文本数据中提取意义,应用范围已经从简单的图像识别扩展到了复杂的视频内容分析,从而为理解和预测公众行为及其背后的意识形态提供了更为深入的视角。由于网络传播路径预测依赖信息转发预测,后者体现了用户的个体行为差异,因而我们提出基于机器学习的网络意识形态风险识别的行为传播模型。该系统由信息采集层、数据分析层、预警发布层构成,可以提高对网络意识形态发展演变状况的实时感知能力,使相关主体能及时完成对网络意识形态风险的评估和决策,推动解决网络意识形态风险的智能化处置。In the era of new media,the rapid development of the internet presents significant challenges to Chi-na’s ideological security and governance.Modern machine learning techniques,particularly those based on deep neural networks,have the ability to extract meaning from vast amounts of unstructured textual data.Their appli-cations have expanded from simple image recognition to complex video content analysis,providing deeper in-sights into understanding and predicting public behavior and underlying ideologies.Given that predicting network dissemination paths relies on forecasting information forwarding,which reflects individual user behavior differ-ences,we propose a behavioral dissemination model for identifying online ideological risks based on machine learning.This system consists of an information collection layer,a data analysis layer,and an early warning lay-er.It enhances real-time awareness of the development and evolution of online ideologies,enabling relevant stakeholders to promptly assess and make decisions regarding ideological risks.This advances the intelligent management of online ideological risks.

关 键 词：机器学习 网络空间 意识形态 数据分析 风险管理 

分 类 号：D64[政治法律—政治学]