A Model for Analyzing Digital Traces in Secure Information and Educational Systems

The paper proposes a model for analyzing digital traces of users in information and educational systems of the university environment in order to increase the level of information security. The proposed model is grounded on a combination of behavioral profile clustering, probabilistic modeling based on Bayesian networks, and machine learning algorithms for risk prediction. The novelty of the model lies in the integration of these methods into a single multilevel structure that provides context-dependent dynamic risk assessment and decision support for secure access to university digital resources. Unlike known solutions, the introduced model allows to take into account the heterogeneity of user behavior, identify hidden dependencies between actions and security events, and adapt to changes in behavioral patterns. Formalization of the key stages of digital trace processing has been conducted. The diagram of model functioning has been created. The approaches to combined calculation of integrated risk based on probabilistic and empirical analysis have been defined. Obtained results can be used as a methodological and technological basis for the development of information security support systems in higher education institutions.