A Modified K-Nearest Neighbor Classifier for E-Mail Spam Detection

Abstract

Electronic mail (e-mail) is one of the most effective and common communication and information exchange methods used to promote products and services. Irrespective of the many benefits e-mail spam offers, it still poses a significant threat in the current Internet ecosystem, leading to loss of revenue by organizations, as well as insecurity and privacy threats to individual users. Therefore, several methods have been devised to counter and reduce spam, especially e-mail classification and filtering methods. Nevertheless, a number of the current solutions have proven to be limited in their ability to differentiate between valid e-mails and spam messages. To overcome these limitations, the paper suggests a customized K-Nearest Neighbor (KNN) algorithm to improve the performance of spam detection. The experimental findings show that the refined KNN algorithm is more accurate in identifying and classifying legitimate and spam e-mails than the traditional methods