Validation of IT Risk Assessments with Markov Logic Networks

Risk assessments of big and complex IT infrastructures comprise numerous qualitative risk estimations for infrastructure assets. Qualitative risk estimations, however, are subjective and thus prone to errors. We present an approach to detect anomalies in the result of risk assessments by considering information about inter-dependencies between various building blocks of IT landscapes from enterprise architecture management. We therefore integrate data from enterprise architecture and risk estimations using Semantic Web technologies and formalize common anomalies such as inconsistent estimations of dependent infrastructure components. To reflect the uncertainty of qualitative analyses we utilize Markov logic networks (MLN) to validate the resulting model and determine more probable and consistent estimations.