DEVELOPMENT OF AN EFFECTIVE SYSTEM FOR DETECTING CYBERCRIMES USING MODIFIED RIPPLE DOWN RULE SYSTEM AND NEURAL NETWORK.

Cybercrime is an unlawful act in which computer is the tools to commit an offense; cyber criminals perform operation in cyber space with the help of the internet. Most existing techniques used in detecting cybercrimes could detect individual attacks but failed in terms of coordinated and distributed attacks. Also, most of the detection system used to curb cybercrimes on web application generates a large number of false alarms. Hence, this research developed an enhanced system which could not only detect individual, coordinated and distributed attacks but also reduce the number of false alarms. The research data for this work which consists of six cards (labeled A, B, C, D, E and F) were sourced from an online shopping store. The six cards contain four attributes with associated two thousand seven hundred (2700) transactions. The number of transactions carried out through each card were 200, 300, 400, 500, 600 and 700 respectively. Sixty percent of transactions carried out on each card were used to train the system while the remaining forty percent were used to test the system. The acquired attributes through each card were used as inputs in developing the system. Radial basis function was used for features extraction and the extracted features were moved to the Modified Ripple Down Rule engine that compared the profiling of the cardholder transaction information. The developed system was implemented on Matrix laboratory environment. The performance of the developed system was evaluated at 0.80 threshold using Sensitivity, Specificity, False Alarm Rate, Accuracy and Computational Time.