Support Vector Machine Based Spam SMS Detection

Short message service (sms) is the most important communication tool in recent decades. with the increased popularity of mobile devices, the usage rate of sms will increase more and more in years. sms is a practical method used to reach individuals directly. but this practical and easy method can cause sms to be misused. the advertising or promotional sms of the companies are an examples of this misuse. in this study, a spam sms detection technique is proposed using data mining (dm) methods. in the proposed study, data mining algorithms such as naive bayes (nb), k-nearest neighborhood (knn), support vector machine (svm), random forest (rf) and random tree (rt) is selected. smsspamcollection dataset, which is contain 747 spam sms and 4827 ham sms, is used. 10 fold cross-validation technique is used to evaluate prediction of spam sms in the dataset. therefore, proposed study achieved 98.33 % success rate and 0,087 false positive rate for svm algorithm.