 |
Software agents are often cited
as a key enabling technology for the next-generation
of online services, such as large-scale electronic commerce
(Guilfoyle, Jeffcoate and Stark, 1997) and Service-Oriented
Computing (Papazoglou, 2003). My current research seeks
to address two key problems with the use of agents within
these services. Firstly, in order to be effective agents
will often need to reason about what is to be done,
i.e. perform practical reasoning (Girle, Hitchcock,
McBurney and Verheij, 2004), but in the proposed services
consisting of large-scale open multi-agent systems,
classical logic-based approaches to reasoning and decision-making
are often unsuitable (Amgoud et al, 2005). Secondly,
key characteristics of the proposed services include
agent heterogeneity, conflicting individual goals, limited
trust and a high probability of non-conformance to specifications
(Artikis, Pitt, and Sergot, 2002). Therefore, if this
vision of large-scale open multi-agent systems is to
be realised then the fundamental problem of interoperability
(i.e. communication between agents) must be addressed. |
In both of these problems I believe
that agents may benefit from the use of argumentation,
a process based on the exchange and valuation of interacting
arguments, to support the process of practical reasoning
and interaction.
Argue
tuProlog
 In
(Bryant, Krause and Vreeswijk, 2006) we discuss our
current work on a light-weight Java-based argumentation
engine (“Argue tuProlog”) that can be used
to implement a non-monotonic reasoning component in
Internet or agent-based applications. The core engine
has been built using tuProlog (Denti, Omicini, and Ricci,
2005. Available: http://www.alice.unibo.it:8080/tuProlog/)
an existing open-source Prolog engine, as its foundation,
which followed the same design principles that we require
for our intended domain of application. Although our
ultimate goal is to create a general purpose argumentation
engine that can be configured to conform to one of a
range of semantics, the current version of the engine
implements the argumentation-based framework presented
in (Amgoud et al, 2006) (allowing our engine to generate
arguments and counter arguments over an inconsistent
knowledge base, determine the acceptability of arguments
and construct proofs using an argument game approach
to constructing proofs of acceptance), and also standard
Prolog inference (allowing us to prototype a variety
of metainterpretters that support other forms of argumentation.)
The motivation behind this strand of research is primarily
to illustrate that a practical Internet ready/agent-based
implementation of argumentation is now viable, and the
fundamental message is that we take seriously the need
for sound empirical evidence for the applicability of
argumentation.
An
Implementation of a Dialogue Manager
There has been much work on agent communication
languages (ACLs), and an increasing amount of this work
has concentrated on argumentation-based dialogue (Amgoud
et al, 2006). However, for an ACL to truly be an enabling
technology, it must rely on a standard or protocol to
ensure that different implementations preserve the ACL's
meaning , and in order to gain acceptance, particularly
for sensitive applications such as electronic commerce,
it must be possible to determine whether or not any
system that claims to conform to an ACL protocol actually
does so (Singh, 1998).
In (Bryant, Krause and Moschoyiannis,
2006) we present a prototype of a tool that demonstrates
how existing limitations in ensuring an agent's compliance
to an argumentation-based dialogue protocol can be overcome.
Dialogue protocols are enforced by means of a series
of distributed "Dialogue Manager" enforcement
components, implemented as a lightweight Java-based
agent proxy. Our ultimate goal with this strand of research
is to implement a generic argumentation-based ACL enforcement
tool, and although currently deployed within a LAN environment
we believe that our system can act as a proof-of-concept
for future large-scale demonstrations of agents engaged
in argumentation-based dialogue, particularly for use
within service-oriented architectures
The above work was partially supported
by the EU IST/STReP ASPIC project, Grant 002307, and
is currently being supported by an EPSRC PhD Studentship.
References
L. Amgoud, M. Caminada, S. Doutre, H. Prakken, and
G. Vreeswijk. Draft formal semantics for ASPIC system.
Technical Report ASPIC Deliverable 2.5, 2005.
L. Amgoud, M. Caminada, P. McBurney, H. Prakken, and
G. Vreeswijk. Final Review and Report on Argumentation
System. Technical Report ASPIC Deliverable 2.6, 2006.
A. Artikis, J. Pitt, and M. Sergot. Animated specifications
of computational societies. In AAMAS '02: Proceedings
of the first international joint conference on Autonomous
agents and multiagent systems, pages 1053-1061, New
York, NY, USA, 2002. ACM Press.
D. Bryant, P. Krause, and G. Vreeswijk, G. ‘Argue
tuProlog: A Lightweight Argumentation Engine for Agent
Applications'. In Proceedings of 1st International Conference
on Computational Models of Argument (COMMA06). Liverpool,
UK. IOS Press. 2006. To appear.
D. Bryant, P. Krause, and S. Moschoyiannis 'A Tool to
Facilitate Agent Deliberation'. In Proceedings of 10th
European Conference on Logics in Artificial Intelligence
(JELIA06), volume 4160 of Lecture Notes in Artificial
Intelligence. Springer. 2006. In press.
E. Denti, A. Omicini, and A. Ricci. Multi-paradigm java-prolog
integration in tuProlog. Sci.Comput. Program., 57(2):217–250,
2005.
R. Girle, D. Hitchcock, P. McBurney and B. Verheij.
Decision support for practical reasoning. In C. C. Reed
and T. J. Norman, editors, Argumentation Machines, pages
56–83. Kluwer Academic Publishers, The Netherlands,
2004.
C. Guilfoyle, J. Jeffcoate, and H. Stark. Agents on
the Web: Catalyst for E-Commerce. Ovun Ltd. London,
1997.
M. P. Papazoglou. Service-Oriented Computing: Concepts,
characteristics and directions. In WISE ’03: Proceedings
of the Fourth International Conference on Web Information
Systems Engineering, page 3, Washington, DC, USA, 2003.
IEEE Computer Society.
M. P. Singh. Agent communication languages: Rethinking
the principles. IEEE Computer, 31(12):40-47, 1998.
|
