No abstract

No abstract

Majority of the present-day organizations store and use data on large scale for their functioning. Often this data is private and essential, and unauthorized access to the data can have serious financial and legal repercussions. While it is easier to capture malicious activities perpetrated by an external agent, it is relatively harder to detect transactions committed by company employee with some malicious intent. This is due to the awareness of the employee about the company database structure along with their authorized access privileges. We propose a novel intrusion detection system called user and role-level cluster-based intrusion detection system (URCIDS). We analyze the user behavior at role and user level, to check if the transaction under consideration is in accordance to the regular behavior of the user. We flag transactions that violate the general access pattern followed by the user. A detailed experimental analysis shows that we are able to gain a higher accuracy than most of the state-of-the-art methodologies proposed in this field of study.

No abstract