multi-layered software architecture

Combining a One-Class Support Vector Machine (OCSVM) with a Convolutional Neural Network (CNN) is presented as a novel technique for detecting fraud in Enterprise Resource Planning (ERP) systems.

The objective of this research is to develop a technique for detecting fraud in Enterprise Resource Planning (ERP) systems by combining a One-Class Support Vector Machine (OCSVM) with a Convolutional Neural Network (CNN), suitable for the Spor Istanbul ERP system.

This study examines the ERP system utilized by Spor Istanbul, the largest sports enterprise in Turkey, as a case study. The study utilizes a custom database of web-based program files to create a dataset of benign and malicious JavaScript applications. Firstly, the text and control flow graph of the program is analyzed. Secondly, the OCSVM method is applied as an outlier detection technique, and CNN is used as a classifier.

The experimental results indicate that the proposed OCSVM-CNN approach achieves higher accuracy (96.78%) in detecting malicious scripts compared to using only CNN (94.8%).

The research contributes to the development of multi-layered ERP software architecture with AI decision support, improving fraud detection in ERP systems.