Knowledge Representation, Scientific Argumentation and Non-monotonic Logic

Models developed by the knowledge representation and reasoning community permit us to study defeasible inference based on argumentation and data. Scientific reasoning progresses by evaluating scientific hypotheses based on data and meta-evidence. Meta-evidence can be understood as arguments for discounting or even ignoring data or other meta-evidence. Non-monotonic reasoning is underpinned by non-monotonic logic. We here develop a method for modelling scientific inferences within formal argumentation models. We show how these models capture hypothesis testing, meta-analysis, strong inference and non-monotonic consequence relations.