摘要
针对在线网络信息内容安全事件的分类问题,利用网络用户通信信息中含有的时间、关系和内容特征均可基于文本描述的特点,引入LDA模型,提出了一种实时多维信息联合(RMIA-LDA)的在线信息内容安全事件分类模型及算法。以网络通信中的时间特征为轴,对由此划分出的各个时间片段中的通信关系、通信内容特征采用LDA模型进行建模分类,对分类结果的相似性进行度量后,再与增量更新数据部分的分类结果归纳合并,从而得到当前实时在线数据中的事件分类。仿真实验结果表明,该模型和算法可以有效实现网络中信息内容安全事件的在线分类,较现有算法具有优越的性能。
Aiming at the problem of event classification of on-line network information content security incidents,LDA was introduced, which uses the characteristics that network user's communication information includes time,relationship and content characters. A realtime multi-dimension information association-LDA( RMIA-LDA) model and algorithm were proposed,which uses the communication time character as timeline,and LDA model to calculate the results of classification of communication relationship and communication content characters in each time slice. After the similarity metric calculation of the classification results and merging them with the results of update data,the result of current on-line events classification was obtained. Experimental results showed that the proposed model and algorithm can achieve the goal of on line information content security incidents classification effectively. It has better performance than the existing algorithms.
出处
《四川大学学报（工程科学版）》
EI
CAS
CSCD
北大核心
2014年第3期70-79,共10页
Journal of Sichuan University (Engineering Science Edition)
基金
国家高技术研究发展计划资助项目(2011AA010605)
国家科技重大专项课题项目(2012ZX03006002-010)
关键词
事件分类
信息内容安全事件
隐含狄利克雷分布
相似性度量
GIBBS抽样
event classification
information content security incidents(ICSI)
latent Dirichlet allocation(LDA)
similarity metric
Gibbs sampling
