Spam Mail Filtering Technique using Different Decision Tree Classifiers through Data Mining Approach - A Comparative Performance Analysis

Abstract

In recent years the highestdegree of communication happens through e-mails which are often affected by passive or active attacks. Effective spam filtering measures are the timely requirement to handle such attacks. Many efficient spam filters are available now-a-days with different degrees of performance and usually the accuracy level varies between 60-80% on an average. But most of the filtering techniques are unable to handle frequent changing scenario of spam mails adopted by the spammers over the time. Therefore improved spam control algorithms or enhancing the efficiency of various existing data mining algorithms to its fullest extent are the utmost requirement.In this paper three types of decision tree classifying techniques which are basically data mining classifiers namely Naïve Bayes Tree classifier (NBT), C 4.5 (or J48) decision tree classifier and Logistic Model Tree classifier (LMT) are studied and analyzed for spam mail filtration. The test results depict that LMT is giving the most efficient result in terms of performance with almost 90% accuracy level to detect spam mails and non-spam (HAM) mails.

Impact of Feature Selection Technique on Email Classification

“…The algorithms based on decision trees have also been used for the purpose of spam filtering. In [8], the author compared three decision tree classifiers namely, Naï ve Bayes Tree (NBT), J48 and Logistic Model Tree (LMT) and it is shown that LMT performed best in terms of accuracy and false positive rate. J48 turned out to be the best classifier in terms of training time.…”

A New Machine Learning based Approach for Text Spam Filtering Technique

“…The most effective and formal method of communication in current days is Electronic mail commonly known as "E mail". More than 500 million people in the world have internet access and the popularity of email technology has grown rapidly in recent years [7]. But, the day to day increase in the number of spam mails has caused a big reason of dissatisfaction amongst the users.…”

“…There are many solutions to spam filtering, e.g., the blacklist and white-list filtering techniques [14], decision tree based approaches [7], [8], [9] and machine learning based methods [4], [15]. Among various solutions, machine learning based ones are receiving more attention due to its high accuracy rate for spam detection.…”

Email Classification Using Machine Learning Algorithms

“…Naïve Bayes Tree classifier, J48 Decision Tree classifier and LMT were analysed. In terms of performance and accuracy level LMT showed best results [2].…”

“…Several researchers have analysed the efficiency of various machine learning algorithms in spam email filtering approaches. The papers [2], [3], [13] and [18] evaluated different classifiers in correctly classifying spam mails. The use of Enron corpus in researches regarding spam filtering was discussed in paper [15].…”