An adaptive spam filter based on Bayesian model and strong features

For the filtering of spam comments on the network, this paper proposes a classification method based on Bayesian model. We first use some generalization patterns to extract such strong features as contact information, which can be easily added to the vector representation of comments. Then, this paper introduces a self-feedback learning mechanism for the filtering of new spam. Through iterative training, it can continually improve the adaptability of the classifier. The experiment on large scale data shows that our method gets a better performance compared with the traditional classifier.