This paper presented a malicious information filtering model based on one-class classification,where one of the classes is sampled very well,while the other class is severely undersampled.

• 针对网络不良信息过滤中训练样本分布不均衡,负面样本对于训练效果无充分贡献,提出了一种利用正面训练样本建立分类模型,并基于该模型实现网络不良信息过滤的方法。