Dung's work "On the
Acceptability of
Arguments and its Fundamental Role in Non-monotonic Reasoning, Logic
Programming and N-persons games" (Artificial Intelligence, 1995) has
provided argumentation theory with the necessary mathematical
abstractions for it to reach scientific maturity, and become an
independent research area within the field of non-monotonic reasoning.
However, while Dung's formal semantics have soon become part of the
basic techniques of argumentation theorists, his argument-based
analysis of N-persons games and logic programming has received
considerably less attention.
The symposium brings together leading
researchers working at the interface of the three disciplines of game
theory, logic and argumentation, in order to foster the interaction
between these research fields along the lines already presented in
Dung's seminal work. A report of the symposium (to appear on
The Reasoner) can be found here
below:
"The GALP symposium held by the
Individual and Collective Reasoning Group (ICS) of the University of
Luxembourg on 23th-24th April 2009 has brought together, for two days,
a number of distinguished researchers who are contributing and have
contributed to interdisciplinary research at the interface of the
disciplines of games, argumentation and logic (with particular focus on
logic programming). The aim of the symposium was to foster the
interaction between the aforementioned research areas along the lines
already present in the seminal contribution of Dung ("On the
Acceptability of Arguments and Its Fundamental Role in Non-Monotonic
Reasoning, Logic Programming, and N-Persons Games", Artificial
Intelligence 1995). While this contribution has laid the foundations of
argumentation theory as a mathematical discipline, sparkling a rich and
lively research area within Artificial Intelligence, its interaction
with Game Theory and Logic Programming has been relatively neglected.
The symposium has filled this gap by highlighting a number of recent
scientific developments as well as stimulating future research
directions.
The talks presented can be grouped
according to four focus points: talks concerning argumentation theory
in general; and talks focusing on the three overlapping areas of games
and logic (programming), games and argumentation, argumentation and
logic.
Argumentation. Dr. Martin Caminada
(University of Luxembourg) has provided a thorough introduction to
argumentation theory presenting novel results concerning, in
particular, algorithmic aspects of argumentation theory and dialogue
games. The implementation of the algorithms introduced by Dr. Caminada
has then been presented in a comprehensive demo by Patrizio Barbini
(Universities of Turin and Luxembourg) and Yining Wu (University of
Luxembourg). Finally, Prof. Gerhard Brewka (University of Leipzig) has
proposed a multi-agent framework for argumentation generalizing Dung's
setting to cover the interaction of different argumentative contexts.
Games and Logic. The contribution of
Prof. Juergen Dix (Technical University of Clausthal) concerned the use
of logic as a formal language for talking about games. It illustrated a
number of systematic extensions of ATL---focusing in particular on
their complexity---able to capture several game-theoretic concepts,
from the typical "power-view" of games based on effectivity functions,
to the full-fledged characterization of equilibrium concepts such as
the Nash equilibrium. Along a similar line, Dr. Marina de Vos
(University of Bath) has shown how Answer Set Programming can be
successively used to encode games and, consequently, compute their Nash
equilibria. Then, somehow closing the circle, she has shown how the
solutions of answer set programs can be seen as the product of playing
winning strategies in appropriately designed logic games.
Argumentation and Games. This
was definitely the richest section in the symposium. Its talks focused
on two main aspects: 1) the game-theoretic proof theory of
argumentation based on the so-called dialogue or discussion games; 2)
the application of argumentation theory to strategic situations in
rational interaction, such as dispute resolution.
In the first group, Dr. Sanjay Modgil
(King's College London) has introduced dialogue games for an extension
of argumentation frameworks incorporating, besides the standard attack
relation between arguments, an attack relation from arguments to attack
relations. Prof. Henry Prakken (Universities of Utrecht and Groningen)
emphasized the procedural and goal-driven aspects of dialogue games,
besides their logical and argumentation-theoretic nature, which still
await a full-fledged formal analysis.
As to the second group, Serena
Villata has proposed an argumentation-theoretic approach to study the
dynamics of coalition-formation in multi-agent systems. Finally,
professor P. M. Dung (Asian Institute of Technology) has introduced a
novel argumentation-theoretic perspective to dispute resolution based
on a form of mechanism design for dialogue games. According to this
perspective dialogue games are viewed as procedures for dispute
resolution where all arguments defensible via the procedure are also
admissible (soundness) and, vice versa, all admissible arguments are
defensible via the procedure (completeness).
Argumentation and Logic. The
symposium hosted two talks which bridged argumentation theory with
modal logic. The first one, by Prof. Dov Gabbay (King's College London)
applied Provability Logic to characterize the content of an
argumentation framework as a modal formula whose models naturally
correspond to the possible complete extensions of the framework. The
second one, by Dr. Davide Grossi (University of Amsterdam) has
systematically investigated the simple idea of viewing Dung's
argumentation frameworks as Kripke models. The talk has shown how such
perspective opens up the possibility of importing techniques (e.g.,
calculi, evaluation games) and results (e.g., complexity of
model-checking) from modal logic to argumentation theory.
All in all, the symposium has
beautifully shown how rich the overlaps are between game theory,
argumentation theory and logic, and how promising future research lines
can be in further investigating such overlaps. For the abstract of the
talks, as well as the slides, see the program page."
Davide Grossi
Institute of Logic, Language and Computation
University of Amsterdam