多媒体网络不良信息过滤方法仿真  

Multimedia Network Adverse Information Filtering Method Simulation

作　　者：宁琳[1] 孙艳红[1] NING Lin;SUN Yan-hong(Library of Chongqing Jiaotong University,Chongqing 400074,Chin)

机构地区：[1]重庆交通大学图书馆,重庆400074

出　　处：《计算机仿真》2018年第7期343-346,共4页Computer Simulation

摘　　要：为了减少多媒体网络的负载,为大众提供一个健康、具有高度利用价值的网络,需要对多媒体网络不良信息进行过滤。当前基于贝叶斯网络的多媒体网络不良信息过滤方法,利用不良信息之间的相互关系将其过滤出去,不良信息特征提取结果不明确,且不良信息的检测效果不理想及过滤精度低。为此提出利用量子进行法将提升寻优能力作为目标,通过多媒体网络不良信息的特征属性Fisher比,组建信息特征子集评价函数。依据量子进化法工作程序,构造多媒体网络不良信息特征选择流程,实现不良信息的特征选择,提升不良信息特征提取效果和检测效果。根据多媒体网络信息包的线性相关性,判断其是否为网络中的创新包;假设线性相关包的数目比一个设定阈值大,且符合多媒体网络不良信息的特征,则确定该创新包是不良信息,将警告传送给其它的网络节点,或者直接将存在不良信息的节点进行过滤。仿真表明,上述方法可增强不良信息特征提取和检测效果,过滤精度相比当前方法更具优势。This article focuses on the quantum evolution which aims to improve the ability of optimization. Through the characteristic attribute Fisher ratio of unhealthy information in multimedia network, the evaluation func- tion of information feature subset was established. According to the process program of quantum evolution method, the process of feature selection for unhealthy information in multimedia network was constructed to achieve feature selec- tion of unhealthy information and improve effect of feature extraction and detection for unhealthy information. Accord- ing to the linear dependence of information package in multimedia network, whether it is an innovative package in net- work was determined. Supposing that the number of linear dependence packages was more than a set threshold value, and it met characteristics of unhealthy information in multimedia network, thus the innovation package was the un- healthy information. Finally, the warning was transmitted to other network nodes, or the node with unhealthy informa- tion was filtered directly. Simulation results prove that this method can enhance the effect of feature extraction and de- tect/on for unhealthy information, and the filtering accuracy is better than that of current method.

关 键 词：多媒体网络 不良信息 过滤 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]