A semiring-based framework for the deduction/abduction reasoning in access control with weighted credentials

We present a variant of the Datalog language (we call it Datalog W), which is able to deal with weights on ground facts. The weights are chosen from a semiring algebraic structure. Our goal is to use this language as a semantic foundation for trust-management languages, in order to express trust relationships associated with a preference (e.g., a cost, an uncertainty, a trust or a fuzzy value). We apply Datalog W as the basis to give a uniform semantics to a weighted extension of the RT language family, called RT W. Moreover, we show that we can model the deduction and abduction reasoning with semiring-based soft constraints: deduction can validate or not the access request, while abduction can be used to compute the missing credentials if the access is denied and the level of preference that would grant the access.