MICHAEL WOOLDRIDGE and MICHAEL FISHER Department of Electronic Engineering, Queen Mary & Westfield College
and Department of Computation, Manchester Metropolitan University
AGENT THEORY
M.J.Wooldridge@qmw.ac.uk and M.Fisher@doc.mmu.ac.uk
The 1990s have seen the emergence and rapid growth fo a new paradigm
in computer software development: agent-based systems. Agents
are computer systems that are capable of flexible autonomous action
in unpredictable, typically multi-agent environments. In this
course, we introduce one of the most important traditions in the
theory of agent systems, in which agents and agent societies are
understood by attributing to them attitudes such as beliefs, desires,
capabilities, intentions, and so on.The course is broadly divided
into two parts. Part (i) begins by introducing the idea of agents
as intentional systems: practical decision makers, reasoning from
their beliefs, and desires, to the actions they perform in a multi-agent
society. We go on to show individual attitudes such as belief
can be formalised within a Kripke-style modal logic framework,
and progressively introduce more complex concepts such as ability,
the interaction of knowledge and action, and intention theory.
Part (i) concludes with a longer case study, in which we introduce
the currently influential belief-desire-intention (BDI) model
of agency. In part (ii), we introduce the use of logic as a tool
for designing and implementing agents. We focus, in particular,
on multi-agent programming languages in which agents directly
execute logical specifications of their intended behaviour. The
course will conclude with a short survey of current and open issues
in agent theory.
The course will assume no prior knowledge of agent theory of the
(typically) modal logic that are used to formalise the properties
of agents. It will, however, assume some general knowledge of
artificial intelligence (such as the role of knowledge representation).
A basic understanding of classical logic (syntax, semantics, and
proof theory) would be helpful. The course will be primarily of
value to those working in computer science or AI, who wish to
gain an understanding of the theory of agency, but will also be
of interst to philosophers and linguists with an interest in how
the concept of agency is treated in AI and computer science. The
course will be supported by extensive notes.
No specific recommendation