Planeamento
Aulas
Class 1. Theoretical class 1. Presentation
Class 4. Theoretical class 2. Autonomous agents, agent societies, and agent communities
Class 7. Theoretical class 4. Agent communication in FIPA-ACL and FIPA-SL. Questions, Subscriptions. Answers
Agent communication in FIPA-ACL and FIPA-SL. Questions, Subscriptions. Answers.
Class 10. Practical class 4 (pencil & paper). Agent communication: questioning, subscription and answering messages
Pencil and paper practical class on agent communication
Main objective: questioning, subscription and answering messages, above all, those of practical classes 3 and 5
Class 13. Theoretical class 8. Agent communication about actions in FIPA ACL and FIPA SL
Class 16. Theoretical class 9. Application scenario presentation and description
Application scenario presentation (e.g., go to restaurant in Goa)
2. Scenario main interactions
3. Scenario preliminary interactions
4. Agent goals
5. Agent actions
6. Agent knowledge
6.1 - Preprogrammed knowledge
6.2 - Information acquired through the interaction
The roles played by agent platforms
Class 19. Theoretical class 12. Inapt Agency: Implementation of the Partner Discovery scenario actions
Class 21. Pratical class 8 (Lab). Implementation and use of the actions of the partner discovery scenario agents
Implementation of the actions of partner discovery scenario agents
Action-based interactions interactions
Class 24. Practical class 9 (Lab). Action preconditions, effects, and impossibilities
Action preconditions, effects, and impossibilities
Blocks world exercises (no agents; just planning)
Class 27. Theoretical class 18. Analysis of the complete partner discovery scenario with planning agents
Class 32. Theoretical class 22. A production system to control the partner discovery agent
Class 35. Theoretical class 23. Preparation of the evaluation. Second part of the subject matter
Preparation of the evaluation. Second part of the subject matter
Aulas
Class 2. Practical Class 1 (Lab). Inapt Agency Installation
Inapt Agenncy Instalation
1. Virtual Box instalation
2. Downloading of the virtual machine image
3. Configuration of the Autonomous Agents virtual machine
4. Downloading the new version of the inapt.jar; replacement of the inapt.jar on the virtual machine with the new version
HelloWorld agent
Class 5. Theoretical class 3. Agent platform. Inapt Agency
Agent platform: definition (set of services) and practice (set of services + implementation tools)
FIPA Agent platform
Agent platform development
- Platform implementations, FIPA compatibility, Software accessibility (paid vs. freeware / open source)
- Companies
- JADE agent platform
Inapt Agency and the JADE agent platform
Inapt Agency roles and mechanisms with respect to (i) message reception and processing; and (ii) goal directed behavior
Inapt Agents knowledge and actions
Class 8. Theoretical class 5. Agent communication in FIPA-ACL and FIPA-SL. Questions, Subscriptions. Answers
Agent communication in FIPA-ACL and FIPA-SL. Questions, Subscriptions. Answers.
Part 2
Class 11. Practical class 5 (Lab). Interaction with knowledge-based agents (information subscription)
Interaction with knowledge-based agents with special emphasis on subscription
- Use the Dummy Agent to interact with the knowledge-based agent
- Use a script agent to interact with the knowledge based agent
Example scenario in which the information in the knowledge based agent changes with time, e.g., a crypto quotes agent.
Class 14. Practical 6 (pencil and paper). Action-based interation
Pencil and paper class on action-based interaction
All request messages and corresponding possible replies (according to the request protocol) in the circular economy scenario
Class 17. Theoretical class 10. Quasi-formal scenario description. An example
Quasi-formal scenario description. An example, e.g., in the circular economy domain
- Agents and agent roles
- Agent interactions
- Agent actions
- Agent knowledge: pre-programmed knowledge and information acquired through the interaction
Class 20. Theoretical class 13. Inapt Agency: Implementation of the Partner Discovery scenario actions
Inapt Agency: Implementation of the Partner Discovery scenario actions
Part 2.
Class 22. Theoretical class 14. Preparation for the evaluation
Preparing the evaluation relative to the first part of the subject matter
Class 25. Theoretical class 16. Action formal descriptions in the partner discovery scenario
Action formal descriptions in the partner discovery scenario: formal description of all actions in the scenario
Class 28. Theoretical class 19. Post-execution waiting conditions
Post-execution waiting conditions
- How to solve the problem of an agent that receives a certain information amount but considers only a part of it? The need for post-execution waiting conditions (n sent questions, n received replies and holding time), statement_response_action specification (w/ conversation id), and action re-implementation
- Specification: post_execution_waiting(Action, ExecutionID)
- Action re-implementation
- statement_response_action
Test the whole scenario again and check that the described approach works well. This could be an exercise for the students
THE SUMMARIES OF CLASSES 27 AND 29 MAY ALSO HAVE TO BE CHANGED
Class 30. Theoretical class 20. Production system: a goal achievement mechanism
Class 33. Practical class 11 (Lab). Additional exercises about production systems
Class 36. Theoretical class 24. Preparation of the evaluation. First part of the subject matter
Preparation of the evaluation. First part of the subject matter
Aulas
Class 3. Practical Class 2 (Lab). Agent platform testing
Class 6. Practical class 3 (Lab). Simple knowledge-based agents
Simple knowledge-based agents
Knowledge-based agents implementation with Inapt Agency
Script "Agents"
Script agents implementation with Inapt Agency
Students scenario
-Testing the knowledge based agent with the Dummy Agent
- Simple agent interaction with a knowledge based agent and a script "agent" (information acquisition)
Class 9. Theoretical class 6. Agent communication in FIPA-ACL and FIPA-SL. Questions, Subscriptions. Answers
Agent communication in FIPA-ACL and FIPA-SL. Questions, Subscriptions. Answers.
Part 3
Class 12. Theoretical class 7. Agent communication about actions in FIPA ACL and FIPA SL
Agent communication in FIPA ACL and FIPA SL: communication about actiosn
1. ACL messages for talking about actions: request family, failure, refuse, agree, cancel, inform
2. The FIPA request protocol
3. Action terms
Part 1
Class 15. Practical 7. Laboratory about action-based agent interaction
Laboratory about action-based agent interaction. Examples in the circular economy scenario
Class 18. Theoretical class 11. Quasi-formal description of the partner discovery scenario
Quasi-formal scenario description. An example in the partner discovery scenario
- Agents and agent roles
- Agent interactions
- Agent actions
- Agent knowledge: preprogrammed knowledge and information acquired through interaction
Class 23. Theoretical class 15. Agent goals. Goal achievement mechanisms: planning algorithms
Agent goals. Examples from the partner discovery scenario.
Goal achievement mechanisms (special purpose programs, production rules, search algorithms, planning algorithms)
Goal regression planning algorithm (Ivan Bratko)
Class 26. Theoretical class 17. Definition of the high-level predicates used in the preconditions and effects of the partner discovery scenario
Definition of the high-level predicates used in the preconditions and effects of the actions of the partner discovery scenario
These high-level predicates are defined in terms of the dynamic low-level scenario predicates. They are needed to avoid using quantifiers
Class 29. Practical class 10 (Lab). Testing the complete partner discovery scenario with planning agents
Testing the complete partner discovery scenario with planning agents
-Tests will be performed both on a local platform and on a remote platform.
THE SUMMARIES OF CLASSES 27 AND 28 MAY ALSO HAVE TO BE CHANGED
Class 31. Theoretical class 21. A production system for a turn changing game
Class 34. Practical 12 (lab). Testing the complete partner discovery scenario (with a production system)
Testing the complete partner discovery scenario, in which the partner finder agent's goal achievement mechanism is a production rule based system. The other scenario agents are controlled by planning algorithms.
Thus, in the described setting, we are dealing with a rather heterogeneous agent society