A Three-Way Decision Approach to Email Spam Filtering

Many classification techniques used for identifying spam emails, treat spam filtering as a binary classification problem. That is, the incoming email is either spam or non-spam. This treatment is more for mathematical simplicity other than reflecting the true state of nature. In this paper, we introduce a three-way decision approach to spam filtering based on Bayesian decision theory, which provides a more sensible feedback to users for precautionary handling their incoming emails, thereby reduces the chances of misclassification. The main advantage of our approach is that it allows the possibility of rejection, i.e., of refusing to make a decision. The undecided cases must be re-examined by collecting additional information. A loss function is defined to state how costly each action is, a pair of threshold values on the posterior odds ratio is systematically calculated based on the loss function, and the final decision is to select the action for which the overall cost is minimum. Our experimental results show that the new approach reduces the error rate of classifying a legitimate email to spam, and provides better spam precision and weighted accuracy.