Applying Agent Technology

Nicholas R. Jennings and Michael Wooldridge Department of Electronic Engineering Queen Mary and Westfield College University of London, London E1 4NS United Kingdom {N.R.Jennings, M.J.Wooldridge}@qmw.ac.uk

Summary

Agent technology is an exciting and fast moving field of IT which is just starting to migrate from universities and research labs into industrial and commercial applications. Given this state of transition, this tutorial presents a practioner's view of the problems and promises which face those who wish to apply agents. Assuming no prior no knowledge of the field, the tutorial will explain the concepts behind agent-based computing and clearly differentiate it from related fields of endeavour. It will identify the types of applications for which agents are suited (and the types for which they are not!), will present case studies from a number of leading edge agent applications, and it will identify the common pitfalls which those new to the field should avoid.

Aims of the Tutorial

After attending this tutorial, a delegate will:

understand what is meant by the terms agent and agent-based system;
understand how agents are different to other, existing software paradigms (such as objects and expert systems);
understand the types of application for which an agent-based solution is appropriate;
be familiar with several major applications of agent technology, understand what benefits agent technology brought to these applications, and what lessons were learned;
be able to identify the main obstacles and pitfalls in the way of the agent system developer, and understand how to recognise and avoid these pitfalls;
understand the potential of agent technology, as well as the main technological and other challenges this technology faces.

Syllabus

An introduction to agents: What are agents? How are agents different to other software paradigms (objects, expert systems)? Some scenarios illustrating the role of agents. [MJW -- 30 minutes]
Agent application domain characteristics: The types of system that can be built using agent technology -- open systems, complex systems, ubiquitous computing. Software engineering and agent systems -- how agents can be used to improve software engineering practice. [NRJ -- 30 minutes]
Case studies: single agent applications [MJW -- 25 minutes] Multi-agent applications [NRJ -- 50 minutes].
Pitfalls of agent-oriented development: The problems most frequently encountered in agent-oriented development projects identified -- their causes, symptoms, and how to rectify them. Limitations of agent-based solutions discussed. [MJW -- 30 minutes]
The future: The challenges and problems facing the agent community, likely areas of future development activity. [NRJ -- 15 minutes]

Pre-Requisites

The tutorial will be at an introductory level, and will assume no prior knowledge of agents or AI. A basic understanding of software development would be helpful, but is not essential.