Comparison between Typical Discriminative Learning Model and Generative Model in Chinese Short Messages Service Spam Filtering

We used the experience of spam filtering on account of Chinese short messages service spam filtering and compared the performances of typical discriminative learning model and generative model, namely naive bayesian model and logistic regression model. Overall, in Chinese short messages service spam filtering, the performance of naive bayesian model is better than logistic regression model using 1-ROCA as evaluating indicator while the final performance of logistic regression model is better than naive bayesian model with the increase in amount of short messages, which is deferent from spam filtering as shown in this experimental results.