基于LVQ2神经网络及决策归纳的中文邮件过滤

Filtering Chinese Spam-mail Based on LVQ2 Neural Network and Decision Tree Classifying

垃圾邮件所带来的巨大经济损失及巨大的信息处理量已成为世界性的问题。目前,国际上应用效果较好的过滤技术是基于贝叶斯概率模型,但是由于汉语极为复杂的语义环境和贝叶斯算法的易欺骗性,使得其不能很好地过滤中文垃圾邮件。对此,该文提出了一种综合LVQ2神经网络及决策树分类的过滤算法。实验结果表明,该算法可以过滤98%以上的中文垃圾邮件。

The great economic losses as well as the arduous work on information processing brought about by spam-mail becomes a world-wide problem. At present, some effective ways of filtering spam-mail are based on Bayes model. However, the complexity of Chinese and the decertfulness of Bayes algorithm make it difficult to filter Chinese spam-mail. In this case, a new algorithm based on LVQ2 neural network and decision tree classifying is put forward. The result of experiment proves that this algorithm can filter over 98% of the Chinese spam-mail.