Probabilistic ABox Reasoning: Preliminary Results

Most probabilistic extensions of description logics focus on the terminological apparatus. While some allow for expressing probabilistic knowledge about concept assertions, systems which can express probabilistic knowledge about role assertions have received very little attention as yet. We present a system PALC which allows us to express degrees of belief in concept and role assertions for individuals. We introduce syntax and semantics for PALC and we define the corresponding reasoning problem. An independence assumption regarding the assertions for different individuals yields additional constraints on the possible interpretations. This considerably reduces the solution space of the PALC reasoning problem.