Effect of N-Grams Technique in Preprocessing of Email Spam Filtering

In this paper, the process of spam categorization based on character level; content-based approach has been demonstrated. Spam categorization has been performed by using N-gram technique. The general technique of using N-grams on words, creating a “Bag of Words” representation of documents, has been replaced by ‘Bag of Characters'.‘Bag of Character' is created by treating the whole email document as a single string and splitting it character-wise. In this approach, multiple N-grams i.e. bi-grams, tri-grams and quad-grams have been used simultaneously. It results in ‘bag of character' representation of email documents containing N-grams of sizes 2, 3 and 4. It enhances the results by enabling us to solve the problems occurring in Word N-grams. All the experiments have been performed on Ling Spam Corpus.