计算机工程

• 人工智能及识别技术 • 上一篇    下一篇

Twitter加密网络行为自动识别方法

朱贺军1,2,祝烈煌2   

  1. (1.北京亿赛通网络安全技术有限公司,北京 100085; 2.北京理工大学计算机学院,北京 100081)
  • 收稿日期:2015-04-30 出版日期:2015-12-15 发布日期:2015-12-15
  • 作者简介:朱贺军(1975-),男,高级工程师、博士研究生,主研方向:大数据处理与挖掘,网络安全;祝烈煌,教授、博士。

Automatic Identification Method of Twitter Encryption Network Behavior

ZHU Hejun  1,2,ZHU Liehuang  2   

  1. (1.Beijing Esafenet Network Security Technology Co., Ltd.,Beijing 100085,China; 2.School of Computer Science & Technology,Beijing Institute of Technology,Beijing 100081,China)
  • Received:2015-04-30 Online:2015-12-15 Published:2015-12-15

摘要: 以Twitter加密数据为研究对象,提出一种快速自动识别加密网络行为的方法。在分析海量Twitter加密网络行为的基础上,提取能够表征加密网络行为的特征,构建加密网络行为模型库。计算实时采集的网络交互数据与模型库中参考样本的相关系数,根据相关系数阈值 进行海量Twitter加密网络行为的自动分类识别。实验结果表明,与基于IP、机器学习等的识别方法相比,该方法能实现Twitter加密网络行为的快速在线自动识别,并解决因加密协议频繁升级导致的开发维护工作量大和在线识别效率低的问题。

关键词: 加密网络行为, 在线识别, 相关系数, 模型库, 数据采集

Abstract: This paper proposes a fast recognition method for the encryption network behavior.The method takes the encrypted Twitter data as the research object,and analyzes a large number of encrypted Twitter network behaviors.Then the characteristics of the encrypted network behavior are extracted,the specific encryption network behavior module database is established,the correlation coefficient between the collection data and the reference sample of module database is calculated,and the encryption network behavior about the mass Twitter is identified automatically by the correlation coefficient threshold. Experimental results show that the method identifies the encryption network behavior quickly and automatically,and reduces the development work when the encryption protocol is upgraded frequently and improves online identification efficiency,compared with the methods using Internet Protocol(IP),machine learning,etc.

Key words: encryption network behavior, online recognition, correlation coefficient, model library, data collection

中图分类号: