摘要
互联网和移动互联网中的信息内容急速膨胀,导致其中充斥着违法违规和不良信息,影响互联网空间的内容安全。基于敏感词匹配的传统文本内容安全识别方法忽略上下文语义,导致误报率高、准确率低。在分析传统文本内容安全识别方法的基础上,提出了利用深度学习的融合识别模型以及模型融合算法流程。深入介绍了基于利用深度学习的融合识别模型的文本内容安全识别系统,并进行了实验验证。结果表明,所提模型可以有效解决传统识别方法缺乏语义理解造成误报率高的问题,提高了不良信息检测的准确性。
The rapid expansion of information content on the internet and the mobile internet has resulted in violations of laws and regulations and bad information, which affects the content security of the internet space. Traditional text content security recognition methods based on matching of sensitive words ignore context semantics, resulting in high false positive rate and low accuracy. Based on the analysis of traditional text content security recognition methods, a fusion recognition model using deep learning and a model fusion algorithm process were proposed. Text content security recognition system based on the fusion recognition model using deep learning and experimental verification was introducted deeply. Results show that the proposed model can effectively solve the problem of high false positive rate caused by the lack of semantic understanding of traditional recognition methods, and improve the accuracy of the bad information detection.
作者
汪少敏
王铮
任华
WANG Shaomin;WANG Zheng;REN Hua(Mobile Intermet System and Application Security National Engineering Laboratory,Shanghai 201315,China;Research Institute of China Telecom Co.,Ltd,Shanghai 200122,China)
出处
《电信科学》
2020年第5期25-30,共6页
Telecommunications Science
关键词
内容安全
违法违规和不良信息
深度学习
文本识别
content security
illegal information and unhealthy information
deep learning
text recognition