Latent Semantic Indexing Based SVM Model for Email Spam Classification

摘要

Internet plays a drastic role in part of communication nowadays but in e-mail, spam is the major problem. Email spam is unwanted, inappropriate or no longer wanted mails also known as junk email. To eliminate these spam mails, spam filtering methods are implemented using classification algorithms. Among various algorithms, Support Vector Machine (SVM) is used as an effective classifier for spam classification by various researchers. But, the accuracy level is not up to notable level so further. To improve the accuracy, Latent Semantic Indexing (LSI) is used as feature extraction method to select the suitable feature space. The hybrid model of spam mail classification can provide the effective results. The Ling spam email corpus is used as datasets for the experimentation. The performance of the system is evaluated using measures such as recall, precision and overall accuracy.