Improved email spam detection model based on support vector machines

Abstract

Email has become extremely popular among people nowadays. In fact, it has been reported to be the cheapest, popular and fastest means of communication in recent times. Despite the huge benefits of emails, unfortunately its usage has been bedeviled with the huge presence of unsolicited and sometimes fraudulent emails which must be promptly detected and isolated through what is popularly referred to as spam detection system. Spam detection is highly needed to protect email users and prevents several negative usages to which emails have been subjected to of recent. Unfortunately, due to the adaptive nature of unsolicited emails through the use of mailing tools, the effectiveness of the spam detecting tools has often been limited and sometimes rendered ineffective, hence the need for better spam detection tools to achieve better spam detection accuracy. Several spam detection models have been proposed and tested in the literature, but still the reported accuracy indicated that there is still need for more work in this direction in order to achieve better accuracy. In this work, support vector machines-based model is proposed for spam detection while paying attention to appropriately search for the optimal parameters to achieve better performance. Experimental results show that the proposed model outperformed all the earlier proposed models on the same popular dataset employed in this work. Accuracy of 95.87 and 94.06% was obtained for training and testing sets, respectively. The 94.06% testing accuracy represents an improvement of 3.11% over the best reported model in the literature that had an accuracy of 91.22% on the same dataset used in this work.