基于K近邻算法的网络敏感信息过滤方法

Network sensitive information filtering method based on K-nearest neighbor algorithm

面对当前方法受到数据稀疏性影响,导致敏感信息过滤效果差的问题,提出了基于K近邻算法的网络敏感信息过滤方法。以K近邻算法中用到的评分实际数据稀疏度为评判依据,对网络敏感信息进行分类,避免了过滤过程受数据稀疏性的影响。构造敏感信息决策树,在树节点上添加敏感关键词,利用K近邻分类算法计算特征值方差矩阵的权重和累积权重。将计算结果添加到反敏感信息库中,引入时间和主题相关度变量参数计算相似度,通过查找网络上的敏感素材,筛选符合条件的敏感信息。由实验结果可知,该方法平均绝对误差和标准化平均绝对误差与其他方法相比数值最小,分别为0.19和0.20,说明其网络敏感信息过滤效果较好。

Facing the problem that the current methods are affected by data sparsity,resulting in poor filtering effect of sensitive information,a network sensitive information filtering method based on K-nearest neighbor algorithm is proposed. Based on the actual data sparsity used in the K-nearest neighbor algorithm,the network sensitive information is classified to avoid the influence of data sparsity on the filtering process. A sensitive information decision tree is constructed,sensitive keywords are added to the tree nodes,and the weight and cumulative weight of eigenvalue variance matrix are calculated by K-nearest neighbor classification algorithm. Add the calculation results to the anti sensitive information database,introduce the time and subject correlation variable parameters,calculate the similarity,and screen the qualified sensitive information by finding the sensitive materials on the network. The experimental results show that the average absolute error and standardized average absolute error of this method are the smallest compared with other methods,which are 0.19 and 0.20 respectively,indicating that the filtering effect of network sensitive information is better.