一种协同过滤算法在网络干扰过滤中的应用  

A Collaborative Filtering Algorithm in the Application of the Network Interference Filter

作　　者：贾忠涛[1] 吴颖川[2] 刘志勤[1] 

机构地区：[1]西南科技大学计算机科学与技术学院,四川绵阳621010 [2]中国空气动力研究与发展中心,四川绵阳621010

出　　处：《计算机仿真》2016年第1期284-287,共4页Computer Simulation

基　　金：国家自然科学基金项目(61202044);四川省科技厅项目(2010GZ0134);四川省教育厅重点项目(09ZA138)

摘　　要：研究网络干扰过滤优化问题。网络通信中的干扰信息对网络性能造成了不容忽视的负面影响,降低了用户搜索目标信息的效率,由于网络干扰信息量大,网络环境复杂,采用传统方法进行过滤时,往往过滤得不够充分,效果较差。为提高通讯精度提出一种基于协同过滤算法的网络干扰过滤方法。依据协同过滤算法理论,构建网络项目的特征相似性模型,为避免评价稀疏产生的不足,采用新颖的相似度计算方法,获取项目之间的相似度,得到项目特征邻居,对项目预测评分进行大小排序,选择排在最前面的N个项作为推荐项给用户,其余的则作为干扰项进行过滤,实现对网络干扰的有效过滤。实验结果表明,采用改进算法进行网络干扰过滤,能够有效地提高过滤的准确度,降低误判率,获得较好的过滤效果,提高了用户获取目标信息的效率,并满足了网络用户的实际需求。Study the network interference filter. The interference information of the network communication of nots allow to ignore the negative effects of network performance, reduce the efficiency of the user search target information, due to large amount of information network interference, network environment, using traditional methods to filter, of- ten enough, filtering effect is poorer, put forward a network disturbance filtering method based on collaborative filte- ring algorithm. Based on the theory of collaborative filtering algorithm, build the network characteristic similarity model of the project, to avoid the sparse resulting shortage, adopting the similarity calculation method, obtaining pro- ject, the similarity between the neighbors get project characteristics, the project size sort prediction score, select the top item as recommended to the user, the rest as interference filter, realize the effective network interference filter. The experimental results show that the improved algorithm for network interference filter, can effectively improve the filtering accuracy, reduce misjudgment rate, achieve better filtering effect and improve the efficiency of the user get target information, and meet the actual needs of Internet users.

关 键 词：协同过滤算法 网络干扰 特征相似度 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]