Spam messages classification algorithm based on BP and isomap

Abstract

In this paper, a classification algorithm for spam messages by using the neural network is proposed. First, the spam messages are pretreated, including word separation, feature word extraction, representation with feature word, and formation of text matrix. Then, the dimensionality of the text matrix is reduced by using isomap algorithm. Finally, the classification is achieved by the BP neural network. According to the experimental results, the algorithm gives good classification results.