A Spam Filtering Technique Using Svmlight Tool

Agrawal Rammohan Ashok, Gaurav Shrivastav

Abstract: Spam has become a headache of the todays internet causing many problems to all users on Internet. Spam emails not only consume computing resources, money but are frustrating. A lot of money is being spent/ lossed due to spam globally each year. Data mining approaches for content based spam filtering have been seen promising. Filtering is the one of the important technique to stop spam. Now-a-days there is a tremendous need of some trust-worthy and adaptive spam filtering system in the market which should have the ability to react quickly to the real time changes and provide fast and qualitative self-tuning in accordance with a new set of features. This paper explores the use of support vector machines for classifying and filtering spam messages and also to improve the accuracy and time performance. Also a comparative analysis among the algorithms is also being presented here with a view to get the optimized results in spam filtering.

Keywords: Data Mining, Machine Learning, Spam Classification, Spam Filtering, Support Vector Machine