Online Journal of Physical and Environmental Science Research

Volume 1 Issue 2, Pages 21-29; June 2012

©2012 Online Research Journals

Available Online at http://www.onlineresearchjournals.org/OJPESR

Full Length Research Article

 

Spam Detection Using Artificial Neural Networks (Perceptron Learning Rule)   Owen Kufandirimbwa*1 and Richard Gotora2   1,2Department of Computer Science, Faculty of Sciences, University of Zimbabwe, Harare, Zimbabwe.   1E-mail: kufandirimbwa@gmail.com; Tel: +263712784287. 2Email: rgotora@gmail.com   Downloaded 30 April, 2012 Accepted 30 May, 2012

Spam is email sent in bulk where there is no direct agreement in place between the recipient and the sender to receive email solicitation. To prevent the delivery of this so called spam, an automated tool called a spam filter is used to recognize spam. The circular nature of these definitions along with their appeal to the intent of sender and recipient make them difficult to formalize. The spam problem seems to persist and the current state of the art techniques in fighting this problem seems not to provide full proof. There are several approaches which try to stop or reduce the huge amount of spam on individuals. These approaches include legislative measures such as anti-spam laws over world-wide. Other techniques are known as Origin-Based filters which are based on using network information and IP addresses in order to detect whether a message is spam or not. The most common techniques are the filtering techniques attempting to identify whether a message is spam or not based on the content and other characteristics of the message. In this paper, we present a technique to spam filtering using Artificial Neural Networks, and the perceptron learning rule.   Keywords: Artificial neural network, spam filtering, perceptron learning rule, training algorithm, learning rate.