Using Feature Selection to Speed Up Online SVM Based Spam Filtering
Yuewu Shen, Guanglu Sun, Haoliang Qi, Xiaoning He · 2010
In this paper, we propose a feature selection method to speed up online SVM based spam filter. Online SVM gives state-of-the-art classification performance on online spam filtering on large benchmark data sets. However, its computational cost is very expensive for large-scale applications. Feature Selection is a crucial step to online SVM classification. We use a feature selection method based on Bayesian reasoning in this paper, and it based on n-gram feature extraction. The Feature Selection method can reduce feature vector dimension and improve the filter performance a little. It can greatly reduce the computational cost of Online SVMs based spam filter. Experimental results show that the feature selection method outperforms pure online SVM for large-scale spam filtering.