摘要
随着互联网的普及,藏文网络空间也面临着日益增多的敏感信息传播风险,给社会稳定和国家安全带来挑战。传统的敏感信息检测方法难以有效应对藏文语言的特殊性和网络信息的复杂性。为了解决这一问题,提出了一种基于CINO-DPCNN的混合神经网络模型。该模型结合了CINO模型对藏文语义的深层次理解和DPCNN模型对文本特征的高效提取能力,能够更准确地识别藏文网络敏感信息。实验结果表明,CINO-DPCNN模型在准确率、F_(1)值等指标方面取得了良好的结果,相较于现有模型有显著提高。这为构建安全、健康的藏文网络环境提供了新的技术支撑,也为其他少数民族语言的敏感信息检测提供了借鉴。
With the increasing popularity of the Internet,the Tibetan-language online space is facing growing risks of sensitive information dissemination,posing challenges to social stability and national security.Traditional methods for detecting sensitive information are unable to effectively address the unique characteristics of the Tibetan language and the complexity of online information.To address this issue,this paper proposes a hybrid neural network model based on CINO-DPCNN.This model combines the deep understanding of Tibetan semantics provided by the CINO model with the high-efficiency feature extraction capabilities of the DPCNN model,enabling more accurate identification of sensitive information in Tibetan-language online networks.The experimental results demonstrate that the CINO-DPCNN model has achieved excellent performance in terms of accuracy,F_(1)score,and other indicators,showing significant improvements over existing models.This provides new technical support for building a secure and healthy Tibetan-language online environment and serves as a reference for sensitive information detection in other minority languages.
作者
吴瑜
严李强
徐梓恒
卓玛央金
Wu Yu;Yan Liqiang;Xu Ziheng;Zhuoma Yangjin(School of Information Science and Technology,Xizang University,Lasa 850000,China)
出处
《网络安全与数据治理》
2025年第4期79-83,共5页
CYBER SECURITY AND DATA GOVERNANCE
基金
国家自然科学基金项目(62406256)
西藏大学研究生高水平人才培养计划项目(2022-GSP-S105)。
作者简介
吴瑜(1999-),男,硕士研究生,主要研究方向:信号与信息处理;通信作者:严李强(1980-),男,硕士,教授,主要研究方向:信号与信息处理。E-mail:158201730@qq.com;徐梓恒(2004-),男,本科生,主要研究方向:藏文信息化。
