A study on the performances of danger theory and negative selection algorithms for mobile spam detection

Spamming activities using text messages on mobile phone are widely spreading, as in line with the development of technology for mobile phones. This phenomenon has contributed a major threat that impacts the usability of messages. Even though many techniques have been proposed and introduced for detecting these ‘unwanted’ messages, all those efforts still cannot bring this problem to an end. The major challenges in detecting and filtering spam messages today are ineffective solution to deal with strains of spam messages because of the variety content of messages and the attitude of users themselves. This paper aims to view the performance of Artificial Immune System (AIS) algorithms inspired from the ideology of Biology Immune System (BIS) in human body for detecting spam messages on mobile phone. Two types of AIS algorithms were used; Danger Theory (DT) and Negative Selection (NS). Their performances were measured and compared in terms of effectiveness, efficiency and Receiver Over Characteristic (ROC) area, tested on WEKA using three different datasets. From our conduction of experiments, generic Negative Selection algorithm performs bette