Intrusion Detection of Masqueraders Based On Data Mining and Soft Computing Techniques

Many organizations face the critical threat of inside attacks from masqueraders who can be either disgruntled employees or external hackers by exploit legitimate user identity to manipulate the system. Intrusion detection systems (IDSs) are deployed to build the normal user profiles and then detect the possible deviation from the past behavior patterns indicating a possible illegal access. In this paper, we apply a profiling method based on user command sequences and apply the data mining technique Naïve Bayes classification to measure the degree of deviation. A fuzzy system is applied to integrate multiple commands execution to evaluate the overall threat of the possible masquerader existence.