Intrusion detection system (IDS) is becoming an integral part of the network security infrastructure. Data mining tools are widely used for developing IDS. There is a lack of researches in the temporal data mining analysis of the intrusions (intrusions detection over different time periods). Most of researches are focusing on the latest snapshot data mining of intrusion detection systems. This work presented in this paper proposes a new temporal data mining analysis technique of intrusion detection systems based on naïve Bayes networks. The presented system considered the time dimension and built many different classifier models to obtain an accurate analysis of intruders. The obtained results give more focusing and deep understanding of the intruders' behavior during the different time periods and illustrate the shrinking and expansions of intruders' classes over the time slices (the migrations of intruders from one segment to another), The temporal analysis of intruders can help in taking an appropriate decision against specific type of attacks (decisions must be suitable with the intruder behaviour). The results indicate the reduction of the possible high positive false rate. |
