Novel Technique of Extraction of Principal Situational Factors for NSSA

The research on Network Security Situational Awareness has become hot area because of increase in reliance on computer networks. The variety of services being provided on the networks has increased many folds. Major problem in this field is to perceive the security situation of the network because of large volume of data produced per unit time, even in a moderate size network. Though the computing capacities of modern machines have increased but to perceive the security situation, very heavy real time data is to processed, which has become a challenge even for modern computing facilities. In this paper data preprocessing technique based feature selection has been proposed. Features reduction is performed using chisquare attribute evaluation and ranker search method. To ascertain the classification performance using reduced feature set Bayesnet and Naivebayes classifiers are used. Current study uses KDD Cup 1999 Train+ data sets as experimental data and comes to conclusion that better situation perception may be achieved by using a small subset of the attributes of dataset. The members of the selected dataset may then be used as situational factors for further analysis of security situation.