Title :
Classification of Turkish spam e-mails with artificial immune system
Author :
Ozdemir, C. ; Atas, M. ; Ozer, A.B.
Author_Institution :
Bilgisayar Muhendisligi Bolumu, Siirt Univ., Siirt, Turkey
Abstract :
In this study, it is aimed to detect frequently encountered spam e-mails with artificial immune algorithms. Turkish spam and non-spam e-mail dataset are generated within the scope of the work. Fisher discriminant analysis (FDA) and Euclidean Distance (ED) are utilized in order to extract features from the turkish email dataset. In order to evaluate the classification accuracies, artificial immune algorithms with Bayes as a linear and artificial neural network as a non-linear classifiers are used. Various artificial immune algorithms, including AIRS1, AIRS2, AIRS2PARALLEL, CLONALG and CSCA are investigated. Among them, CSCA reveals the best classification accuracy of 86%. Furthermore, CSCA algorithm classifies spam emails with 81% and non-spam e-mails with 90% accuracies.
Keywords :
Bayes methods; Internet; artificial immune systems; computational geometry; natural language processing; neural nets; pattern classification; statistical analysis; unsolicited e-mail; AIRS1 algorithm; AIRS2 algorithm; AIRS2PARALLEL algorithm; CLONALG algorithm; CSCA algorithm; Euclidean distance; FDA; Fisher discriminant analysis; Internet; Turkish email dataset; Turkish spam e-mail classification; artificial immune algorithms; artificial immune system; artificial neural network; feature extraction; linear classifiers; nonlinear classifiers; nonspam e-mail dataset; spam e-mail detection; Algorithm design and analysis; Classification algorithms; FAA; Feature extraction; Internet; Unsolicited electronic mail; Create a dataset; Turkish spam e-mails; artificial immune algorithms; csca; fisher;
Conference_Titel :
Signal Processing and Communications Applications Conference (SIU), 2013 21st
Conference_Location :
Haspolat
Print_ISBN :
978-1-4673-5562-9
Electronic_ISBN :
978-1-4673-5561-2
DOI :
10.1109/SIU.2013.6531457
