Multi-Agent Systems

1
Multi-Agent Systems E-Commerce

• Martin Beer,
• School of Computing Management Sciences,
• Sheffield Hallam University, Sheffield,
• United Kingdom
• m.beer_at_shu.ac.uk

2
Objectives

• To look at how multi-agent services can be used
  to construct online trading communities
• To understand the technology needed to manage
  such services
• Theoretical basis
• Multi-Agent Communication
• Negotiating Strategies
• Platform Infrastructure Issues
• To gain some practical experience of deploying
  multi-agent applications

3
Administrative Issues

• Course materials (slides, tutorial papers, book
  lists etc) will be loaded into Blackboard
• This will become the course repository and will
  include links to software, book sites etc. as
  they become relevant
• I shall upload materials a few days in advance of
  their use
• This gives us flexibility to develop the course
  in directions you want
• Allows latest developments to be incorporated as
  we go along
• Discourages forward browsing

4
Timetable
Lecture One per week (Tuesday 1000)
Tutorials (Alternate Weeks Friday 900)
Practicals (Alternate Weeks Tuesday 400)
5
Course Structure

• Designed to be Flexible
• More general issues first technical detail
  later
• Causes an issue with practicals
• Tried to split assessments so that you do not
  become overloaded
• We need to look at some of the theory but I
  shall try not to examine it (ie exam questions
  will ask for descriptions understanding, not
  mathematics)
• I have spent the summer on agent binfrastructure
  issues. If I get bogged down in infrastructure
  Rescue me!

6
Course Structure First Semester
Lecture Tutorial Practical
Week 1 Introduction No Tutorial or Practical No Tutorial or Practical
Week 2 What are Agents? Nwana Introduction to Agents Paper A simple Agent Application
Week 3 Characteristics of Agents Nwana Introduction to Agents Paper A simple Agent Application
Week 4 Robocup Nwana Introduction to Agents Paper Introduction to Robocup
Week 5 Information Agents Nwana Introduction to Agents Paper Introduction to Robocup
Week 6 Brokering Jennings Wooldridge Selecting Playing Teams
Week 7 Personal Mobile Agents Jennings Wooldridge Selecting Playing Teams
Week 8 Cooperating Agents Jennings Wooldridge Selecting Playing Teams
Week 9 Cooperating Agents Jennings Wooldridge Selecting Playing Teams
Week 10 Transaction Management Jasper Paper Selecting Playing Teams
Week 11 Agents for Financial Apps Jasper Paper Selecting Playing Teams
Week 12 Agents for Financial Apps Reserve Reserve
7
Course Structure Second Semester
Lecture Tutorial Practical
Week 13 Agent Architectures Zeus Paper Building a simple Multiagent System
Week 14 Formal Definition of Agents (Not Examined) Zeus Paper Building a simple Multiagent System
Week 15 Formal Definition of Agents (Not Examined) Zeus Paper Building a simple Multiagent System
Week 16 Building Agent Communities Zeus Paper Building a simple Multiagent System
Week 17 Agent Communication Languages Jade Paper Building a simple Multiagent System
Week 18 Agent Communication Languages Jade Paper Building a simple Multiagent System
Week 19 Negotiation Jade Paper Building a simple Multiagent System
Week 20 Negotiation Jade Paper Building a simple Multiagent System
Week 21 Auctions Sample Healthcare System (Honeywell) Building a simple Multiagent System
Week 22 Auctions Sample Healthcare System (Honeywell) Building a simple Multiagent System
Week 23 Summary Reserve Reserve
Week 24 Reserve Reserve Reserve
8
Assessment
Assessment Assessment Assessment
Tutorial Assignment 25 Wednesday of Week 12 (Noon)
Practical Assignment 25 Wednesday of Week 24 (Noon)
Examination 50
9
Teaching Materials

• Books
• Will circulate book list in a couple of weeks
• No books ideal
• Most very theoretical
• Technology changing so fast that authors unable
  to keep up
• Little real practical knowledge
• Papers
• More useful but incomplete
• Tutorial papers chosen to build up into a useful
  learning resource
• Will not totally replace a text book
• But will go a long way towards it

10
Practical Work

• We shall use a number of agent tools to gain some
  practical experience
• Toolkits based on Java
• Jade
• Zeus
• Links already installed on Blackboard
• RoboCup simulator will be used to demonstrate
  some principles

11
When will agents be implemented? (Trudeau, 1996)