Towards subjective networks: Extending conditional reasoning in subjective logic

This work expands the concept of conditional reasoning in subjective logic between two variables to reasoning over multiple variables as a step to eventually realize efficient means for inference over subjective networks, i.e. Bayesian networks with uncertain marginal and conditional probabilities. Three new inference methods are introduced for a three node subjective network. Monte Carlo simulations characterize the accuracy of these methods to represent inferred uncertain knowledge about the ground truth.