Title: Argument Schemes and Critical Questions for Decision Aiding Process
Speaker: Wassila Ouerdane, LAMSADE, Université Paris Dauphine
Date: Monday, April 21, 2008 12:00 - 1:30 pm
Location: DyDan Center, CoRE Bldg, Room 431, Rutgers University, Busch Campus, Piscataway, NJ
Abstract:
Decision-support systems aims at helping the user to shape a problem situation, formulate a problem and possibly establish a viable solution to it. Under such a perspective decision support can be seen as the construction of the reason for which an action is considered a "solution to a problem" rather than the solution itself. Indeed the problem of "accountability" of decisions is almost as important as the decision itself. Decision support can therefore be seen as an activity aiming to construct arguments through which a decision maker will convince first herself and then other actors involved in problem situation that "that action" is the best one. Decision Theory and Multiple Criteria Decision Analysis have focussed on these issues for a long time, but more on how this "best solution" should be established and less on how a decision maker should be convinced about that.
More recently, in the field of Artificial Intelligence, Argumentation Theory has been put forward as a very general approach allowing to support different kinds of decision-making. Then, our ambition is to try to specify in argumentative terms some of the reasoning steps involved in a decision-aiding process. To do that, we make use of the popular notion of argument schemes, and specify the related critical questions. As we shall see, a hierarchical structure of argument schemes allows to decompose the process into several distinct steps?and for each of them the underlying premises are made explicit, which allows in turn to identify how these reasoning steps can be dialectically defeated via critical questions. This work initiates a systematic study which aims at constituting a significant step forward for forthcoming decision-aiding tools. The kind of system that we foresee and sketch here would allow: (i) to present a recommendation that can be explicitly justified; (ii) to revise any piece of reasoning involved in this process, and be informed of the consequences of such moves; and possibly (iii) to stimulate the client by generating contradictory arguments.