Design and Development of Antispammer for SMS Spam Detection

Abstract

The growth of the mobile phone users has led to a dramatic increase in SMS spam messages. Though in most parts of the world, mobile messaging channel is currently regarded as “clean” and trusted, on the contrast recent reports clearly indicate that the volume of mobile phone spam is dramatically increasing year by year. The success of the mobile messaging channel has, unfortunately, made it a very attractive target for attack by spammers. It is an extremely growing problem, primarily due to the availability of very cheap bulk pre-pay SMS packages and the fact that SMS engenders higher response rates as it is a trusted and personal service. Here trust means almost all the messages received by the subscribers are opened and read at least once. Also because of the ease of use of Smartphones, numbers are easily dialled or links can be smoothly clicked, exposing the subscriber to more risk. To further exacerbate, the situation attackers are finding the traditional fixed email channel increasingly unprofitable and are focusing their activities on the SMS channel. The growing volume of spam messages has increased the demand for accurate and efficient spam solutions. SMS spam filtering is a relatively new task which inherits many issues and solutions from email spam filtering. Many spam solutions have been proposed in the recent past. The one which we address in this thesis, treats spam detection as a simple two class document classification problem. The solution will consist of classification algorithm coupled with feature extractions. Classification along with appropriate features helped us improving the performance in terms of accuracy and has lesser computational time and storage requirements. In this dissertation, we compared the performance achieved by several established machine learning techniques with our approaches. Additionally, we present details about a real and public SMS spam collection based on the perspective of Indian hams and spams. Further, it was analysed with various classifiers for the best results.