A Fast Methodology to Find Decisively Strong Association Rules (DSR) by Mining Datasets of Security Records

Cybersecurity bulletins officially recognize and publicly share the vulnerabilities of Information Systems. The attacks exploit various aspects of those vulnerabilities, compromising confidentiality, integrity or availability of the data collected. We analyze a public dataset of security records so to obtain some common features and to be able to forecast future attacks. We propose an intervention based on history of attacks through data mining methods and so a more dynamic risk analysis, by concentrating on some specific classes of cyberattacks in a period of two years. We devise a fast algorithm to find strong rules which provide an estimate of the probability that these attacks will occur so to identify adequate controls and countermeasures.