Naive bayes for web penetration behavior classification based on word segmentation improvement

After using naive bayes algorithm to mark, train and classify the web log, it is not good to find out the web penetration behavior. In this paper, a parametric word segmentation algorithm suitable for web penetration behavior classification is proposed by analyzing the characteristics of URL parameters and perfecting the operation process of naive bayes. It is proved that this algorithm can effectively reflects the behavior features and greatly improve the classification performance of naive bayes by application and contrast.