“…In this article, we are still concerned with intentional hazards (i.e., those related to malevolent acts) and address issue (iii) above, i.e., the definition of the actions to undertake for reducing the level of system vulnerability. In particular, the empirical classification model developed in Wang et al (2013) is tailored to address the corresponding inverse (classification) problem (Aggarwal, Chen, & Han, 2006, 2010Ahuja & Orlin, 2001, 2002Heuberger, 2004;Li, Zhou, & Zhang, 2012;Lin, Kuo, Hsieh, & Wang, 2009;Mannino & Koushik, 2000;Mousseau,Özpeynirci, &Özpeynirci, 2018;Mousseau & Slowinski, 1998;Pendharkar, 2002), i.e., the problem of determining a set of protective actions Larsson, 1992), which can effectively reduce the vulnerability class of (a group of) safety-critical systems (Aven & Flage, 2009), taking into account a specified set of constraints (e.g., budget limits; Aggarwal et al, 2010).…”