Email classification for Spam Detection using Word Stemming

Abstract

Unsolicited emails, known as spam, are one of the fast growing and costly problems associated with the Internet today. Among the many proposed solutions, a technique using Bayesian filtering is considered as the most effective weapon against spam. Bayesian filtering works by evaluating the probability of different words appearing in legitimate and spam mails and then classifying them based on that probabilities.Most of the current spam email detection systems use keywords to detect spam emails.These keywords can be written as misspellings eg: baank or bannk instead of bank. Misspellings are changed from time to time and hence spam email detection system needs to constantly update the blacklist to detect spam emails containing misspellings. It"s impossible to predict all possible misspellings for a given keyword and add those to the blacklist. In this paper a better and more successful approach for improving E-mail content classification for spam control is proposed. It used the Word Stemming or Word Hashing Technique for improving the efficiency of the content based spam filter.The proposed system extract the base or stem of a misspelled or modified word, to detect spam emails. It considers every misspelled keyword applies a word stemming technique and passes the base word to the content based filter. Using a proposed if-then rule, we can decide whether or not this unknown mail is spam [1].This paper also provides an Email archiving solution which classifies the E-mail relating to a person,

Detection of fraudulent emails by employing advanced feature abundance

“…The authors used J48 algorithm in order to formulate rules to generate concepts of the ontology. The study by Renuka and Hamsapriya [15] adapted the use of word stemming instead of simply content based words for spam email detection. The authors showed that stemming based method is more efficient as compared to content based methods.…”

Perbandingan Metode Random Forest dan Naïve Bayes dalam Email Spam Filtering

“…Hasil dari penelitian sebelumya yang dilakukan oleh Renuka & Hamsapriya [1] menggunakan pengaplikasian word stemming akurasi dari metode Naïve Bayesian dapat ditingkatkan sampai 5-10% dimana metode naïve bayes lebih baik daripada metode random forest. Sedangkan penelitian menggunakan metode random forest yang dilakukan oleh Akinyelu & Adewumi [2] mendapatkan nilai presisi yang cukup tinggi dibanding metode naïve bayes dan tingkat kesalahan yang rendah.…”

Detecting spam e-mails using stop word TF-IDF and stemming algorithm with Naïve Bayes classifier on the multicore GPU

“…Rusland et al [5] worked on enhancing the conventional spam detection technique which uses Naïve Bayes algorithm for classification of spam mails. Renuka and Hamsapriya [9] analysed spam filtration. Bayesian filter works by testing the probability of various words appearing in legitimate, valid and spam mails and then classifying them based on those probabilities as spam or not [9].…”

“…Renuka and Hamsapriya [9] analysed spam filtration. Bayesian filter works by testing the probability of various words appearing in legitimate, valid and spam mails and then classifying them based on those probabilities as spam or not [9]. Shabbir and Mithun [3] showed that if some sort of word stemming or word hashing technique is used that can extract the base or stem of a misspelled or modified word, then the efficiency of any content based spam filter can be significantly improved [10].…”