基于特征提取的垃圾邮件检测

SPAM DETECTING BASED ON FEATURE EXTRACTION ALGORITHM

垃圾邮件处理作为一种典型的文本分类应用问题,受到高维数据的困扰。为提高垃圾邮件检测的效率和准确率,提出一种基于PLS特征提取和SVM的入侵检测算法,首先对原始垃圾邮件数据利用偏最小二乘算法降低维度,再采用遗传算法寻优转换特征子集,并通过支持向量机SVM进行分类。Matlab仿真实验表明,本算法能有效降低数据维数,提高检测的准确率。

As a typical text classification application problem, spam detecting is confused by the high dimensional data problems. In order to improve the efficiency and accuracy of spam detection, this paper proposes an intrusion detection algorithm based on PLS and SVM. The original spam data is projected through using the partial least squares algorithm to reduce the dimension of feature extraction and the genetic algorithm to find the best presented features, and the data is classified by the support vector machine. The Matlab simulation experiments show that our methods can effectively reduce the dimension of data and improve the accuracy of detecting.