Content based hybrid sms spam filtering system

Abstract

World has changed. Everybody is connected. Almost each and everyone have a mobile phone. Millions of SMSs are going around the world over mobile networks in every second. But about 113 of them are spam. SMS spam has become a crucial problem with the increase of mobile penetration around the world. SMS spam filtering is a relatively new task which inherits many issues and solutions from email spam filtering. However it poses its own specific challenges. Server based approaches and Mobile application based approaches are accommodate content based and content less mechanism to do the SMS spam filtering. Though there are approaches, still there is a lack of a hybrid solution which can do general filtering at server level while user specific filtering can be done on mobile level. This paper presents a hybrid solution for SMS spam filtering where both feature phone users as well as smart phone users get benefited. Feature phone users can experience the general filter while smart phone users can configure and filter SMSs based on their own preferences rather than sticking in to a general filter. Server level solution consists of a neural network along with a Bayesian filter and device level filter consists of a Bayesian filter. We have evaluated the accuracy of neural network using spam huge dataset along with some randomly used personal SMSs.