A Predictive, Context-Dependent Stochastic Model for Engineering Applications

This work explores the architecture of a context-dependent probabilistic model. We identify opportunities for providing reminders to operators in their environment as a means to address information overload. Hence, there is a need to represent a state of knowledge and help them stay vigilant during their jobs. Along with the architectural improvements, which further specialize information flows and develop a data-driven approach, continual learning techniques covered events in a probabilistic graphical model called Context-Dependent Recommendation Systems (CD-RS). We demonstrated, as a result, the use of statistical thinking and Design of Experiments (DoE), which are most clear in conducting a suitable experiment. Moreover, the validation of the model and experiments of the novel architecture based on the collected data from a real case study demonstrates the value of the proposed methods.