Performance Analysis of Email Classifiers for Detection of Spam

Abstract

Growing usage of email has also increased size of email data, this data involves important as well as undesirable emails. Amount of unwanted emails(spam) has increased enormously. Blocking spam sources doesn’t works well in this era. For saving resources its vital to separate spam and essential emails(ham). Email servers are prepared to tackle this situation. Problem is handled by different algorithms that automate the system instead of manually separating emails. Our work addresses the selection of algorithm, whose outcome will precisely allocate labels to emails and will be efficient enough to give results in adequate time. So, that emails can be classified correctly into inbox and spam folders in adequate time by email server. Three different machine learning classifiers are analyzed over a dataset, providing a criterion that will categorize them according to their time, precision, recall and accuracy.