Management of Uncertain Data in Event Graphs

We consider graphs to model uncertain facts as edges, linking involved entities, with weights reflecting uncertainty degree. Rules are used to create new edges from the existing ones, and methods to propagate uncertainty measures are defined using a suitable theoretical framework. We also consider new rules, mined from graphs containing uncertain information and answer sets obtained using such rules. We then use Argument Graphs and Possibility Networks to evaluate the conclusions that can be drawn from the facts, taking into account their uncertainty. Finally, information revision is discussed for cases when a new piece of information is added to the graph.