Start Date
11-12-2016 12:00 AM
Description
Fueled by digitization, IT ubiquity and increased connectedness in today's corporations, the need for increased database security becomes evident. A case in point are threats originating from malicious insiders. Effective intrusion detection systems are needed to protect company interests as well as customer privacy from insider attacks. Current detection systems mainly focus on the protection of database systems in stable environments. However, fast-growing databases as well as complex requirements to database roles require dynamic and more flexible intrusion detection approaches. We apply design-oriented research to improve upon current intrusion detection systems. To this end, we first illustrate the design requirements for database protection in dynamic environments using a brief example. We then describe our artifact which extends the misuseability-score approach to facilitate dynamic updating of sensitivity levels using weighted Naive-Bayes classification. We conclude by outlining our future research steps concerning implementation and evaluation.
Recommended Citation
Rappel, Niklas; Stein, Nikolai; and Flath, Christoph, "Dynamic Intrusion Detection in Database Systems: A Machine-Learning Approach" (2016). ICIS 2016 Proceedings. 8.
https://aisel.aisnet.org/icis2016/ISSecurity/Presentations/8
Dynamic Intrusion Detection in Database Systems: A Machine-Learning Approach
Fueled by digitization, IT ubiquity and increased connectedness in today's corporations, the need for increased database security becomes evident. A case in point are threats originating from malicious insiders. Effective intrusion detection systems are needed to protect company interests as well as customer privacy from insider attacks. Current detection systems mainly focus on the protection of database systems in stable environments. However, fast-growing databases as well as complex requirements to database roles require dynamic and more flexible intrusion detection approaches. We apply design-oriented research to improve upon current intrusion detection systems. To this end, we first illustrate the design requirements for database protection in dynamic environments using a brief example. We then describe our artifact which extends the misuseability-score approach to facilitate dynamic updating of sensitivity levels using weighted Naive-Bayes classification. We conclude by outlining our future research steps concerning implementation and evaluation.