Abstract—The paper presents a suspicious email detection model which incorporates enhanced feature selection. In the paper we proposed the use of feature selection strategies along with classification technique for terrorists email detection. The presented model focuses on the evaluation of machine learning algorithms such as decision tree (ID3), logistic regression, Naïve Bayes (NB), and Support Vector Machine (SVM) for detecting emails containing suspicious content. In the literature, various algorithms achieved good accuracy for the desired task. However, the results achieved by those algorithms can be further improved by using appropriate feature selection mechanisms. We have identified the use of a specific feature selection scheme that improves the performance of the existing algorithms.
Index Terms—Decision tree, feature selection, logistic regression, naive bayes, SVM.
S. Nizamani, N. Memon, and U. Kock are with at Maersk McKinney Moller Institute, University of Southern, Denmark. (email: firstname.lastname@example.org).
P. Karampelas is with Department of Hellenic American University, Manchester, NH, USA (email: email@example.com).
Cite: Sarwat Nizamani, Nasrullah Memon, Uffe Kock Wiil, and Panagiotis Karampelas, "Modeling Suspicious Email Detection Using Enhanced Feature Selection," International Journal of Modeling and Optimization vol. 2, no. 4, pp. 371-377, 2012.