Advanced hybrid stochastic dynamic Bayesian network inference algorithm development in the context of the web applications test execution

Abstract

The article is devoted to the application of dynamic Bayesian networks models for fuzzing web applications and development of effective hybrid algorithms for probabilistic inference based on particle filter algorithm. Dynamic Bayesian networks models allow to simulate the dynamic process transformation of web applications associated with the process of their constant instrumental and logical updates, and create a probabilistic structure required for learning process of testing the top web applications vulnerabilities, that able to use the evidence and inference results obtained in the retrospective and current testing slices and improve testing mechanisms in new time slices. The hybrid probabilistic inference algorithm for dynamic Bayesian networks models for testing web-applications, proposed in the current research, significantly increase the efficiency of the classical approximate probabilistic inference algorithms, well reflect the features of the temporary testing links formation and adapted to the detection of anomalous errors.