A comprehensive study of spam detection in e-mails using bio-inspired optimization techniques

Abstract

Electronic mail is a medium of communication used frequently for conveying a variety of information. It has become an integral part of people's lives owing to its ease of access and capability to reach out to a large number of recepients without much hassle. This boon, however, has turned into a bane due to exploitation by marketers for publicizing their products, and scammers for fooling people into falling for their schemes. Such e-mails are usually termed as spam e-mails. The menace of spam e-mails requires attention because in addition to consuming resources like bandwidth and storage, they are time-consuming as their removal may require manual effort. This paper proposes a classification method based on k-Nearest Neighbours integrated with several bio-inspired optimization techniques to classify e-mails as spam or legitimate. The study first evaluates the performance of three distance metrics namely Euclidean, Manhattan, and Chebyshev, when utilized in k-Nearest Neighbours classification. Further, five bio-inspired algorithms namely, Grey wolf optimization, Firefly optimization, Chicken swarm optimization, Grasshopper optimization, and Whale optimization, have been explored and their performance is compared based on different evaluation measures like accuracy, precision, recall, speed of convergence to global optimum solution, F1-measure and computational time.