An Improved SVM-KNN Spam Filtering Approach

摘要

A novel feature weighted SVM-KNN spam filtering algorithm is proposed,which trains SYM respectively according to the different lengths of samples feature and acquires different weights of different lengths of samples feature in KNN sample library-.KNN classifies the testing samples with the acquired weights.Experiments demonstrate the new algorithm has better accuracy of spam filtering with lower computational burden compared with general SVM-KNN.