A method of SMS spam filtering based on AdaBoost algorithm

Short message is one of the most common communication media for mobile subscribers, so major mobile operators are devoted to improve their Short Message Service (SMS). However, the annoying and undesired messages, also named message spam or simply spam, not only worsen the users' experience, but also cause their complaints on SMS. In this paper, we present a novel Chinese SMS spam filtering framework based on AdaBoost algorithm to provide accurate and effective short messages classification. Three content-based weak filters are introduced to boost the performance of final classification decision. Results from Receiver Operating Characteristics (ROC) analysis prove the proposed method has such advantages as higher efficiency and fewer parameters over those established SMS spam filtering methods. The application of the proposed method is expected to block the most spam for mobile subscribers and improve the service quality of SMS. With simple data processing and few training parameters, the proposed method can be applied into the practice of short text classification.

I. Introduction

Along with the rapid development of mobile communication technology, short message has become a popular way for mobile subscribers to send message since the 21^th century. In China, between 2005 and 2010, the text message volume per year has grown exponentially (304.65 billion to 831.7 billion), which greatly changed the communication way. Additionally, as a mobile value-added service, Short Message Service (SMS) is enriched and improved by all major mobile operators both in coverage and transmission rate. Unfortunately, huge amount of unsolicited junk messages flood users' in-boxes, which leads to valuable time lost, bandwidth cost and lots of unnecessary trouble [1]. These annoying and undesired messages, also named message spam or simply spam, not only worsen user experience, but also cause complaints on SMS. What's worse, the ease of generating content and decreasing cost of sending spam messages make the problem more and more serious. In recent years, governments worldwide have shown increasing concern on the issue of spam abuse and taken legal, economical and technical measures to solve the problem.