Spam Message Classification Based on the Naieve Bayes Classification Algorithm

摘要

A classification model based on the naive Bayes algorithm is proposed to classify spam messages more effectively. Spam message classification models based on the naive Bayes algorithm are constructed both for multi-classification and multi-two-classification through steps involving text preprocessing based on regular expression and feature extraction based on Jieba segmentation and the TF-IDF (term frequency-inverse document frequency) algorithm. By further comparing the classification performance against the support vector machine and random forest algorithms, the naieve Bayes algorithm based on multi-two-classification is shown to be the best.