An approach to argumentation-based reasoning under uncertainty

Argumentation is an important mechanism for nonmonotonic reasoning. Extending argumentation to permit reasoning about uncertainty would result in reasoners capable of operating in a wide variety of realistic domains. This paper describes one such formalism, which associates uncertainty with abstract argument, thereby allowing us to determine the level of uncertainty with respect to argument justification. We describe an application of this approach within CISpaces, a virtual space for collaborative intelligence analysis among coalitions. Uncertainty is present in the incoming information at the basis of the hypotheses identified during the analysis. By adopting our approach, CISpaces provides support to analysts in identifying how likely the hypotheses are justified.