Deep Learning-Enabled Big Data Analytics for Cybersecurity Threat Detection in ERP Ecosystems

Abstract

We propose to integrate deep learning and big data analytics to better manage cybersecurity threats that may emerge within an enterprise resource planning (ERP) ecosystem. Big data analytics on their own are insufficient for a rapidly growing threat landscape that becomes more sophisticated by the day. Deep learning can assist in analyzing complex patterns and structures in unstructured data. However, very few studies have hitherto combined these two technologies to specifically detect cybersecurity threats in an ERP environment. With this background, this analysis has two major objectives: (a) to investigate ways in which deep learning can be coupled with big data analytics to help contain cybersecurity threats in an ERP ecosystem and (b) to study the kinds of common threats, if any, that are currently present within an ERP environment.

This research is unique in that it presents an exclusive and comprehensive approach to detecting cybersecurity threats that may emerge within the internal processes and activities of an ERP ecosystem, using new and advanced analytical methods. In this study, cybersecurity threats were examined using various firewalls from different organizations located all across the globe, and the common related logs were collected. A new anomaly detection approach that integrates a deep learning technique into a big data analytics setup was initialized in order to check the new attempts and their patterns. The results have shown that the approach presented has the potential to be employed to detect fraud within ERP ecosystems, as well as gather insights into common cybersecurity threats that are generally observed. In other words, the detection of security threat factors is shown to be encouraging; thus, the practical research agenda has been proposed.