A classification modelling for fraud sms identification using text mining

Abstract

Nowadays the presence of fraudulence based on SMS is increasing in the society. Those SMS use unsuspicious sentences, so people could be misled. We propose a classification model which indicates the presence of fraud and use the pattern to predict the new SMS. This research use ROCK algorithm to cluster data and Naive Bayes algorithm to build classifier. Moreover, semantic word correction technique was applied. SMS which have indication of fraud, used in this research are SMS containing request of handphone voucher to certain phone number, offering to be voucher agent, and interest in the activities of buying and selling land. 5 cluster with threshold 0.08 are resulted from clustering phase and result from training data phase is classifier with accuracy 80,93%. Finally, the classifier are successfully implemented on the Android phone.