Argumentation is a heterogeneous concept which has been studied in several fields, such as Rhetoric, Linguistics and Philosophy. In this talk, we will see some of the approaches to argumentation in Artificial Intelligence. In particular, we will discuss the ways in which formal argumentation can be applied to the development and evaluation of models and specifications, specially in domains where being able to represent and generate arguments is a natural approach to formalise reasoning. Examples are domains where different viewpoints need to be considered or where formal proofs cannot always be provided, such as safety-critical systems.
We have been working in an argumentation framework that can capture and represent some of the dynamics involved in an argumentation process over the design of a model. We will describe how we intend to formalise this by using domain specific argument schemas, which are instantiated in order to attack or defend some property or behaviour of a model. In order to illustrate these ideas, we will work through a simple example.