![Page 1: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/1.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Multi-agent Semantic Web Systems:Agent Communication Overview
Jacques Fleuriot
![Page 2: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/2.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
1 Agents Overview
2 Agent ArchitecturesMiddle AgentsNetwork Architectures
3 Agent Communication
4 Interaction Models
5 Summary
![Page 3: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/3.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
What are Agents?
‘Intelligent’ characteristics of agents:
autonomyreasoning abilitylearning abilitymobilitysociabilitycooperationnegotiation
Unlike standard WS, an agent can reason about:
how to handle external requests,what order to carry them out.
![Page 4: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/4.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
What are Agents?
‘Intelligent’ characteristics of agents:
autonomy
reasoning abilitylearning abilitymobilitysociabilitycooperationnegotiation
Unlike standard WS, an agent can reason about:
how to handle external requests,what order to carry them out.
![Page 5: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/5.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
What are Agents?
‘Intelligent’ characteristics of agents:
autonomyreasoning ability
learning abilitymobilitysociabilitycooperationnegotiation
Unlike standard WS, an agent can reason about:
how to handle external requests,what order to carry them out.
![Page 6: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/6.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
What are Agents?
‘Intelligent’ characteristics of agents:
autonomyreasoning abilitylearning ability
mobilitysociabilitycooperationnegotiation
Unlike standard WS, an agent can reason about:
how to handle external requests,what order to carry them out.
![Page 7: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/7.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
What are Agents?
‘Intelligent’ characteristics of agents:
autonomyreasoning abilitylearning abilitymobility
sociabilitycooperationnegotiation
Unlike standard WS, an agent can reason about:
how to handle external requests,what order to carry them out.
![Page 8: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/8.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
What are Agents?
‘Intelligent’ characteristics of agents:
autonomyreasoning abilitylearning abilitymobilitysociability
cooperationnegotiation
Unlike standard WS, an agent can reason about:
how to handle external requests,what order to carry them out.
![Page 9: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/9.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
What are Agents?
‘Intelligent’ characteristics of agents:
autonomyreasoning abilitylearning abilitymobilitysociabilitycooperation
negotiation
Unlike standard WS, an agent can reason about:
how to handle external requests,what order to carry them out.
![Page 10: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/10.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
What are Agents?
‘Intelligent’ characteristics of agents:
autonomyreasoning abilitylearning abilitymobilitysociabilitycooperationnegotiation
Unlike standard WS, an agent can reason about:
how to handle external requests,what order to carry them out.
![Page 11: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/11.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
What are Agents?
‘Intelligent’ characteristics of agents:
autonomyreasoning abilitylearning abilitymobilitysociabilitycooperationnegotiation
Unlike standard WS, an agent can reason about:
how to handle external requests,what order to carry them out.
![Page 12: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/12.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
What are Agents?
‘Intelligent’ characteristics of agents:
autonomyreasoning abilitylearning abilitymobilitysociabilitycooperationnegotiation
Unlike standard WS, an agent can reason about:
how to handle external requests,
what order to carry them out.
![Page 13: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/13.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
What are Agents?
‘Intelligent’ characteristics of agents:
autonomyreasoning abilitylearning abilitymobilitysociabilitycooperationnegotiation
Unlike standard WS, an agent can reason about:
how to handle external requests,what order to carry them out.
![Page 14: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/14.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Multi-Agents Systems (MAS)
Distributed system which incorporates independent agents.
The collection of agents interact, and
solve problems that are outside their individual capacities.
Focus on properties that emerge from cooperation (vs.capabilities of individual agents)
‘standard’ distributed system: coordination must bespecified in advance
Multi-Agent system: (some aspects of) coordinationachieved dynamically at run-time
Seen as complementary to existing Semantic Web/Gridtechnologies.
![Page 15: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/15.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Multi-Agents Systems (MAS)
Distributed system which incorporates independent agents.
The collection of agents interact, and
solve problems that are outside their individual capacities.
Focus on properties that emerge from cooperation (vs.capabilities of individual agents)
‘standard’ distributed system: coordination must bespecified in advance
Multi-Agent system: (some aspects of) coordinationachieved dynamically at run-time
Seen as complementary to existing Semantic Web/Gridtechnologies.
![Page 16: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/16.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Multi-Agents Systems (MAS)
Distributed system which incorporates independent agents.
The collection of agents interact, and
solve problems that are outside their individual capacities.
Focus on properties that emerge from cooperation (vs.capabilities of individual agents)
‘standard’ distributed system: coordination must bespecified in advance
Multi-Agent system: (some aspects of) coordinationachieved dynamically at run-time
Seen as complementary to existing Semantic Web/Gridtechnologies.
![Page 17: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/17.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Multi-Agents Systems (MAS)
Distributed system which incorporates independent agents.
The collection of agents interact, and
solve problems that are outside their individual capacities.
Focus on properties that emerge from cooperation (vs.capabilities of individual agents)
‘standard’ distributed system: coordination must bespecified in advance
Multi-Agent system: (some aspects of) coordinationachieved dynamically at run-time
Seen as complementary to existing Semantic Web/Gridtechnologies.
![Page 18: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/18.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Multi-Agents Systems (MAS)
Distributed system which incorporates independent agents.
The collection of agents interact, and
solve problems that are outside their individual capacities.
Focus on properties that emerge from cooperation (vs.capabilities of individual agents)
‘standard’ distributed system: coordination must bespecified in advance
Multi-Agent system: (some aspects of) coordinationachieved dynamically at run-time
Seen as complementary to existing Semantic Web/Gridtechnologies.
![Page 19: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/19.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Multi-Agents Systems (MAS)
Distributed system which incorporates independent agents.
The collection of agents interact, and
solve problems that are outside their individual capacities.
Focus on properties that emerge from cooperation (vs.capabilities of individual agents)
‘standard’ distributed system: coordination must bespecified in advance
Multi-Agent system: (some aspects of) coordinationachieved dynamically at run-time
Seen as complementary to existing Semantic Web/Gridtechnologies.
![Page 20: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/20.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Multi-Agents Systems (MAS)
Distributed system which incorporates independent agents.
The collection of agents interact, and
solve problems that are outside their individual capacities.
Focus on properties that emerge from cooperation (vs.capabilities of individual agents)
‘standard’ distributed system: coordination must bespecified in advance
Multi-Agent system: (some aspects of) coordinationachieved dynamically at run-time
Seen as complementary to existing Semantic Web/Gridtechnologies.
![Page 21: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/21.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Coordination Problem
Simplistic WS Model:
“one-shot” interactions:client sends a request message to a single service operationand receives a response meessage.
In practice, we want to allow more complex kinds ofinteraction:
multiple operations,multiple messages exchanged,messages sequenced in a particular order,multiple parties involved in the interaction.
How do we ensure that such interactions are
coordinated?correct?robust to failures?
![Page 22: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/22.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Coordination Problem
Simplistic WS Model:
“one-shot” interactions:
client sends a request message to a single service operationand receives a response meessage.
In practice, we want to allow more complex kinds ofinteraction:
multiple operations,multiple messages exchanged,messages sequenced in a particular order,multiple parties involved in the interaction.
How do we ensure that such interactions are
coordinated?correct?robust to failures?
![Page 23: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/23.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Coordination Problem
Simplistic WS Model:
“one-shot” interactions:client sends a request message to a single service operationand receives a response meessage.
In practice, we want to allow more complex kinds ofinteraction:
multiple operations,multiple messages exchanged,messages sequenced in a particular order,multiple parties involved in the interaction.
How do we ensure that such interactions are
coordinated?correct?robust to failures?
![Page 24: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/24.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Coordination Problem
Simplistic WS Model:
“one-shot” interactions:client sends a request message to a single service operationand receives a response meessage.
In practice, we want to allow more complex kinds ofinteraction:
multiple operations,multiple messages exchanged,messages sequenced in a particular order,multiple parties involved in the interaction.
How do we ensure that such interactions are
coordinated?correct?robust to failures?
![Page 25: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/25.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Coordination Problem
Simplistic WS Model:
“one-shot” interactions:client sends a request message to a single service operationand receives a response meessage.
In practice, we want to allow more complex kinds ofinteraction:
multiple operations,
multiple messages exchanged,messages sequenced in a particular order,multiple parties involved in the interaction.
How do we ensure that such interactions are
coordinated?correct?robust to failures?
![Page 26: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/26.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Coordination Problem
Simplistic WS Model:
“one-shot” interactions:client sends a request message to a single service operationand receives a response meessage.
In practice, we want to allow more complex kinds ofinteraction:
multiple operations,multiple messages exchanged,
messages sequenced in a particular order,multiple parties involved in the interaction.
How do we ensure that such interactions are
coordinated?correct?robust to failures?
![Page 27: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/27.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Coordination Problem
Simplistic WS Model:
“one-shot” interactions:client sends a request message to a single service operationand receives a response meessage.
In practice, we want to allow more complex kinds ofinteraction:
multiple operations,multiple messages exchanged,messages sequenced in a particular order,
multiple parties involved in the interaction.
How do we ensure that such interactions are
coordinated?correct?robust to failures?
![Page 28: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/28.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Coordination Problem
Simplistic WS Model:
“one-shot” interactions:client sends a request message to a single service operationand receives a response meessage.
In practice, we want to allow more complex kinds ofinteraction:
multiple operations,multiple messages exchanged,messages sequenced in a particular order,multiple parties involved in the interaction.
How do we ensure that such interactions are
coordinated?correct?robust to failures?
![Page 29: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/29.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Coordination Problem
Simplistic WS Model:
“one-shot” interactions:client sends a request message to a single service operationand receives a response meessage.
In practice, we want to allow more complex kinds ofinteraction:
multiple operations,multiple messages exchanged,messages sequenced in a particular order,multiple parties involved in the interaction.
How do we ensure that such interactions are
coordinated?correct?robust to failures?
![Page 30: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/30.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Coordination Problem
Simplistic WS Model:
“one-shot” interactions:client sends a request message to a single service operationand receives a response meessage.
In practice, we want to allow more complex kinds ofinteraction:
multiple operations,multiple messages exchanged,messages sequenced in a particular order,multiple parties involved in the interaction.
How do we ensure that such interactions are
coordinated?
correct?robust to failures?
![Page 31: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/31.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Coordination Problem
Simplistic WS Model:
“one-shot” interactions:client sends a request message to a single service operationand receives a response meessage.
In practice, we want to allow more complex kinds ofinteraction:
multiple operations,multiple messages exchanged,messages sequenced in a particular order,multiple parties involved in the interaction.
How do we ensure that such interactions are
coordinated?correct?
robust to failures?
![Page 32: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/32.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Coordination Problem
Simplistic WS Model:
“one-shot” interactions:client sends a request message to a single service operationand receives a response meessage.
In practice, we want to allow more complex kinds ofinteraction:
multiple operations,multiple messages exchanged,messages sequenced in a particular order,multiple parties involved in the interaction.
How do we ensure that such interactions are
coordinated?correct?robust to failures?
![Page 33: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/33.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Interaction Example
Client
Service
Login
Search
Order
Pay
Logout
![Page 34: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/34.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Types of Middle Agent
Middle-agents:
assist in locating service providers, andconnecting service providers with service requesters.
Two important types of middle-agent(http://www.cs.cmu.edu/~softagents/middle.html)
Matchmaker (yellow pages agents): receivesadvertisements and matches with requests.
Broker: like matchmaker, but also processes therequests.
![Page 35: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/35.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Types of Middle Agent
Middle-agents:
assist in locating service providers, and
connecting service providers with service requesters.
Two important types of middle-agent(http://www.cs.cmu.edu/~softagents/middle.html)
Matchmaker (yellow pages agents): receivesadvertisements and matches with requests.
Broker: like matchmaker, but also processes therequests.
![Page 36: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/36.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Types of Middle Agent
Middle-agents:
assist in locating service providers, andconnecting service providers with service requesters.
Two important types of middle-agent(http://www.cs.cmu.edu/~softagents/middle.html)
Matchmaker (yellow pages agents): receivesadvertisements and matches with requests.
Broker: like matchmaker, but also processes therequests.
![Page 37: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/37.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Types of Middle Agent
Middle-agents:
assist in locating service providers, andconnecting service providers with service requesters.
Two important types of middle-agent(http://www.cs.cmu.edu/~softagents/middle.html)
Matchmaker (yellow pages agents): receivesadvertisements and matches with requests.
Broker: like matchmaker, but also processes therequests.
![Page 38: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/38.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Types of Middle Agent
Middle-agents:
assist in locating service providers, andconnecting service providers with service requesters.
Two important types of middle-agent(http://www.cs.cmu.edu/~softagents/middle.html)
Matchmaker (yellow pages agents): receivesadvertisements and matches with requests.
Broker: like matchmaker, but also processes therequests.
![Page 39: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/39.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Types of Middle Agent
Middle-agents:
assist in locating service providers, andconnecting service providers with service requesters.
Two important types of middle-agent(http://www.cs.cmu.edu/~softagents/middle.html)
Matchmaker (yellow pages agents): receivesadvertisements and matches with requests.
Broker: like matchmaker, but also processes therequests.
![Page 40: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/40.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Service Matchmaking
Matchmaker
Requestor Provider
Request for Service
Advertise Service
Reply Provider Agents' Names
![Page 41: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/41.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Service Matchmaking
Matchmaker
Requestor Provider
Request for Service
Advertise Service
Reply Result of Service
Reply Provider Agents' Names
Request for Service
![Page 42: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/42.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Service Brokering
Broker
Requestor Provider
Request for ServiceRequest for Service
Advertise Service
Reply Result of ServiceReply Result of Service
![Page 43: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/43.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Network Architectures
So far, mainly assumed some kind of centralizedclient/server architecture.
But Service Oriented Architectures can equally well bedecentralized
![Page 44: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/44.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Network Architectures
So far, mainly assumed some kind of centralizedclient/server architecture.
But Service Oriented Architectures can equally well bedecentralized
![Page 45: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/45.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Centralized: Client/Server
![Page 46: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/46.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Decentralized: Peer-to-Peer
![Page 47: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/47.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Peer-to-Peer Architectures
P2P is...
a self-organizing system of equal, autonomous entities (peers)[which] aims for the shared usage of distributed resources in anetworked environment avoiding central services.
Peers interact directly with each other, usually withoutcentral coordination.
Each peer has autonomy over its own resources.
Within a set of peers, each uses resources provided byother peers.
Peers can act as both clients and servers; i.e., no intrinsicasymmetry of role.
Performance considerations may dictate some centralizedelements in P2P systems — leads to hybrid P2P systems.
![Page 48: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/48.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Peer-to-Peer Architectures
P2P is...
a self-organizing system of equal, autonomous entities (peers)[which] aims for the shared usage of distributed resources in anetworked environment avoiding central services.
Peers interact directly with each other, usually withoutcentral coordination.
Each peer has autonomy over its own resources.
Within a set of peers, each uses resources provided byother peers.
Peers can act as both clients and servers; i.e., no intrinsicasymmetry of role.
Performance considerations may dictate some centralizedelements in P2P systems — leads to hybrid P2P systems.
![Page 49: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/49.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Peer-to-Peer Architectures
P2P is...
a self-organizing system of equal, autonomous entities (peers)[which] aims for the shared usage of distributed resources in anetworked environment avoiding central services.
Peers interact directly with each other, usually withoutcentral coordination.
Each peer has autonomy over its own resources.
Within a set of peers, each uses resources provided byother peers.
Peers can act as both clients and servers; i.e., no intrinsicasymmetry of role.
Performance considerations may dictate some centralizedelements in P2P systems — leads to hybrid P2P systems.
![Page 50: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/50.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Peer-to-Peer Architectures
P2P is...
a self-organizing system of equal, autonomous entities (peers)[which] aims for the shared usage of distributed resources in anetworked environment avoiding central services.
Peers interact directly with each other, usually withoutcentral coordination.
Each peer has autonomy over its own resources.
Within a set of peers, each uses resources provided byother peers.
Peers can act as both clients and servers; i.e., no intrinsicasymmetry of role.
Performance considerations may dictate some centralizedelements in P2P systems — leads to hybrid P2P systems.
![Page 51: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/51.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Peer-to-Peer Architectures
P2P is...
a self-organizing system of equal, autonomous entities (peers)[which] aims for the shared usage of distributed resources in anetworked environment avoiding central services.
Peers interact directly with each other, usually withoutcentral coordination.
Each peer has autonomy over its own resources.
Within a set of peers, each uses resources provided byother peers.
Peers can act as both clients and servers; i.e., no intrinsicasymmetry of role.
Performance considerations may dictate some centralizedelements in P2P systems — leads to hybrid P2P systems.
![Page 52: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/52.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Peer-to-Peer Architectures
P2P is...
a self-organizing system of equal, autonomous entities (peers)[which] aims for the shared usage of distributed resources in anetworked environment avoiding central services.
Peers interact directly with each other, usually withoutcentral coordination.
Each peer has autonomy over its own resources.
Within a set of peers, each uses resources provided byother peers.
Peers can act as both clients and servers; i.e., no intrinsicasymmetry of role.
Performance considerations may dictate some centralizedelements in P2P systems — leads to hybrid P2P systems.
![Page 53: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/53.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Hybrid: Peers and Super-Peers
![Page 54: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/54.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Agent Architectures
MAS can have any of the above architectures.
Middle-agent will play role of coordinator in a centralizedarchitecture.
Increasing interest in achieving coordination in pure P2Psystems.
![Page 55: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/55.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Agent Architectures
MAS can have any of the above architectures.
Middle-agent will play role of coordinator in a centralizedarchitecture.
Increasing interest in achieving coordination in pure P2Psystems.
![Page 56: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/56.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Agent Architectures
MAS can have any of the above architectures.
Middle-agent will play role of coordinator in a centralizedarchitecture.
Increasing interest in achieving coordination in pure P2Psystems.
![Page 57: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/57.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Agent Communication
Communication involves sharing at various levels,including
how to describe states of affairs (vocabulary, grammar),how to carry out speech acts (performative verbs),how to engage in dialogue.
![Page 58: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/58.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Agent Communication
Communication involves sharing at various levels,including
how to describe states of affairs (vocabulary, grammar),
how to carry out speech acts (performative verbs),how to engage in dialogue.
![Page 59: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/59.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Agent Communication
Communication involves sharing at various levels,including
how to describe states of affairs (vocabulary, grammar),how to carry out speech acts (performative verbs),
how to engage in dialogue.
![Page 60: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/60.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Agent Communication
Communication involves sharing at various levels,including
how to describe states of affairs (vocabulary, grammar),how to carry out speech acts (performative verbs),how to engage in dialogue.
![Page 61: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/61.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
DARPA Knowledge Sharing Effort (KSE)
http://www-ksl.stanford.edu/knowledge-sharing/
Project from early ’90s aimed at developing techniques,tools and re-usable resources to support buildinglarge-scale knowledge-based systems and knowledge bases.
Results in the area of knowledge representation:
Syntax KIF — Knowledge Interchange FormatSemantics Ontolingua — language for defining sharable
ontologiesPragmatics KQML — high-level interaction language
![Page 62: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/62.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
DARPA Knowledge Sharing Effort (KSE)
http://www-ksl.stanford.edu/knowledge-sharing/
Project from early ’90s aimed at developing techniques,tools and re-usable resources to support buildinglarge-scale knowledge-based systems and knowledge bases.
Results in the area of knowledge representation:
Syntax KIF — Knowledge Interchange FormatSemantics Ontolingua — language for defining sharable
ontologiesPragmatics KQML — high-level interaction language
![Page 63: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/63.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
DARPA Knowledge Sharing Effort (KSE)
http://www-ksl.stanford.edu/knowledge-sharing/
Project from early ’90s aimed at developing techniques,tools and re-usable resources to support buildinglarge-scale knowledge-based systems and knowledge bases.
Results in the area of knowledge representation:
Syntax KIF — Knowledge Interchange FormatSemantics Ontolingua — language for defining sharable
ontologiesPragmatics KQML — high-level interaction language
![Page 64: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/64.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
DARPA Knowledge Sharing Effort (KSE)
http://www-ksl.stanford.edu/knowledge-sharing/
Project from early ’90s aimed at developing techniques,tools and re-usable resources to support buildinglarge-scale knowledge-based systems and knowledge bases.
Results in the area of knowledge representation:
Syntax KIF — Knowledge Interchange Format
Semantics Ontolingua — language for defining sharableontologies
Pragmatics KQML — high-level interaction language
![Page 65: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/65.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
DARPA Knowledge Sharing Effort (KSE)
http://www-ksl.stanford.edu/knowledge-sharing/
Project from early ’90s aimed at developing techniques,tools and re-usable resources to support buildinglarge-scale knowledge-based systems and knowledge bases.
Results in the area of knowledge representation:
Syntax KIF — Knowledge Interchange FormatSemantics Ontolingua — language for defining sharable
ontologies
Pragmatics KQML — high-level interaction language
![Page 66: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/66.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
DARPA Knowledge Sharing Effort (KSE)
http://www-ksl.stanford.edu/knowledge-sharing/
Project from early ’90s aimed at developing techniques,tools and re-usable resources to support buildinglarge-scale knowledge-based systems and knowledge bases.
Results in the area of knowledge representation:
Syntax KIF — Knowledge Interchange FormatSemantics Ontolingua — language for defining sharable
ontologiesPragmatics KQML — high-level interaction language
![Page 67: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/67.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Knowledge Interchange Format
KIF: First order logic with set theory.
Intended as interlingua for encoding declarative knowledge.
Features:
semantics independent of implementationexpressivenon-decidablehuman readable
Current specification athttp://logic.stanford.edu/kif
![Page 68: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/68.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Knowledge Interchange Format
KIF: First order logic with set theory.
Intended as interlingua for encoding declarative knowledge.
Features:
semantics independent of implementationexpressivenon-decidablehuman readable
Current specification athttp://logic.stanford.edu/kif
![Page 69: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/69.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Knowledge Interchange Format
KIF: First order logic with set theory.
Intended as interlingua for encoding declarative knowledge.
Features:
semantics independent of implementationexpressivenon-decidablehuman readable
Current specification athttp://logic.stanford.edu/kif
![Page 70: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/70.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Knowledge Interchange Format
KIF: First order logic with set theory.
Intended as interlingua for encoding declarative knowledge.
Features:
semantics independent of implementation
expressivenon-decidablehuman readable
Current specification athttp://logic.stanford.edu/kif
![Page 71: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/71.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Knowledge Interchange Format
KIF: First order logic with set theory.
Intended as interlingua for encoding declarative knowledge.
Features:
semantics independent of implementationexpressive
non-decidablehuman readable
Current specification athttp://logic.stanford.edu/kif
![Page 72: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/72.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Knowledge Interchange Format
KIF: First order logic with set theory.
Intended as interlingua for encoding declarative knowledge.
Features:
semantics independent of implementationexpressivenon-decidable
human readable
Current specification athttp://logic.stanford.edu/kif
![Page 73: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/73.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Knowledge Interchange Format
KIF: First order logic with set theory.
Intended as interlingua for encoding declarative knowledge.
Features:
semantics independent of implementationexpressivenon-decidablehuman readable
Current specification athttp://logic.stanford.edu/kif
![Page 74: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/74.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Knowledge Interchange Format
KIF: First order logic with set theory.
Intended as interlingua for encoding declarative knowledge.
Features:
semantics independent of implementationexpressivenon-decidablehuman readable
Current specification athttp://logic.stanford.edu/kif
![Page 75: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/75.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
KQML
Knowledge Query and Manipulation Language (KQML).
High-level, message-oriented language for informationexchange.
Makes no assumptions about:
transport mechanism;content language;ontology used in message content.
Message types particularly oriented towards multi-agentcommunication.
![Page 76: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/76.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
KQML
Knowledge Query and Manipulation Language (KQML).
High-level, message-oriented language for informationexchange.
Makes no assumptions about:
transport mechanism;content language;ontology used in message content.
Message types particularly oriented towards multi-agentcommunication.
![Page 77: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/77.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
KQML
Knowledge Query and Manipulation Language (KQML).
High-level, message-oriented language for informationexchange.
Makes no assumptions about:
transport mechanism;content language;ontology used in message content.
Message types particularly oriented towards multi-agentcommunication.
![Page 78: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/78.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
KQML
Knowledge Query and Manipulation Language (KQML).
High-level, message-oriented language for informationexchange.
Makes no assumptions about:
transport mechanism;
content language;ontology used in message content.
Message types particularly oriented towards multi-agentcommunication.
![Page 79: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/79.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
KQML
Knowledge Query and Manipulation Language (KQML).
High-level, message-oriented language for informationexchange.
Makes no assumptions about:
transport mechanism;content language;
ontology used in message content.
Message types particularly oriented towards multi-agentcommunication.
![Page 80: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/80.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
KQML
Knowledge Query and Manipulation Language (KQML).
High-level, message-oriented language for informationexchange.
Makes no assumptions about:
transport mechanism;content language;ontology used in message content.
Message types particularly oriented towards multi-agentcommunication.
![Page 81: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/81.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
KQML
Knowledge Query and Manipulation Language (KQML).
High-level, message-oriented language for informationexchange.
Makes no assumptions about:
transport mechanism;content language;ontology used in message content.
Message types particularly oriented towards multi-agentcommunication.
![Page 82: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/82.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
KQML Message
Example Message
(tell :sender amazon.com:receiver info-agent0011:in-reply-to msid-7.24.97.45391:ontology http://amazon.com/Books:language prolog:content ‘‘price(item567, gbp, 24.95)’’
)
performative
parameter (an attribute/value pair)
declarative message content
![Page 83: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/83.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
KQML Message
Example Message
(tell :sender amazon.com:receiver info-agent0011:in-reply-to msid-7.24.97.45391:ontology http://amazon.com/Books:language prolog:content ‘‘price(item567, gbp, 24.95)’’
)
performative
parameter (an attribute/value pair)
declarative message content
![Page 84: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/84.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
KQML Message
Example Message
(tell :sender amazon.com:receiver info-agent0011:in-reply-to msid-7.24.97.45391:ontology http://amazon.com/Books:language prolog:content ‘‘price(item567, gbp, 24.95)’’
)
performative
parameter (an attribute/value pair)
declarative message content
![Page 85: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/85.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
KQML Message
Example Message
(tell :sender amazon.com:receiver info-agent0011:in-reply-to msid-7.24.97.45391:ontology http://amazon.com/Books:language prolog:content ‘‘price(item567, gbp, 24.95)’’
)
performative
parameter (an attribute/value pair)
declarative message content
![Page 86: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/86.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
FIPA
FIPA (Foundation for Intelligent Physical Agents;http://www.fipa.org/
Standards organization that promotes agent-basedtechnology, based in Geneva.
An IEEE Computer Society standards committee since2005.
Has specifications in the areas of;
agent applicationsabstract architectureagent communicationagent managementagent message transport
![Page 87: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/87.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
FIPA
FIPA (Foundation for Intelligent Physical Agents;http://www.fipa.org/
Standards organization that promotes agent-basedtechnology, based in Geneva.
An IEEE Computer Society standards committee since2005.
Has specifications in the areas of;
agent applicationsabstract architectureagent communicationagent managementagent message transport
![Page 88: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/88.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
FIPA
FIPA (Foundation for Intelligent Physical Agents;http://www.fipa.org/
Standards organization that promotes agent-basedtechnology, based in Geneva.
An IEEE Computer Society standards committee since2005.
Has specifications in the areas of;
agent applicationsabstract architectureagent communicationagent managementagent message transport
![Page 89: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/89.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
FIPA
FIPA (Foundation for Intelligent Physical Agents;http://www.fipa.org/
Standards organization that promotes agent-basedtechnology, based in Geneva.
An IEEE Computer Society standards committee since2005.
Has specifications in the areas of;
agent applicationsabstract architectureagent communicationagent managementagent message transport
![Page 90: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/90.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
FIPA
FIPA (Foundation for Intelligent Physical Agents;http://www.fipa.org/
Standards organization that promotes agent-basedtechnology, based in Geneva.
An IEEE Computer Society standards committee since2005.
Has specifications in the areas of;
agent applications
abstract architectureagent communicationagent managementagent message transport
![Page 91: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/91.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
FIPA
FIPA (Foundation for Intelligent Physical Agents;http://www.fipa.org/
Standards organization that promotes agent-basedtechnology, based in Geneva.
An IEEE Computer Society standards committee since2005.
Has specifications in the areas of;
agent applicationsabstract architecture
agent communicationagent managementagent message transport
![Page 92: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/92.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
FIPA
FIPA (Foundation for Intelligent Physical Agents;http://www.fipa.org/
Standards organization that promotes agent-basedtechnology, based in Geneva.
An IEEE Computer Society standards committee since2005.
Has specifications in the areas of;
agent applicationsabstract architectureagent communication
agent managementagent message transport
![Page 93: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/93.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
FIPA
FIPA (Foundation for Intelligent Physical Agents;http://www.fipa.org/
Standards organization that promotes agent-basedtechnology, based in Geneva.
An IEEE Computer Society standards committee since2005.
Has specifications in the areas of;
agent applicationsabstract architectureagent communicationagent management
agent message transport
![Page 94: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/94.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
FIPA
FIPA (Foundation for Intelligent Physical Agents;http://www.fipa.org/
Standards organization that promotes agent-basedtechnology, based in Geneva.
An IEEE Computer Society standards committee since2005.
Has specifications in the areas of;
agent applicationsabstract architectureagent communicationagent managementagent message transport
![Page 95: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/95.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
FIPA
FIPA (Foundation for Intelligent Physical Agents;http://www.fipa.org/
Standards organization that promotes agent-basedtechnology, based in Geneva.
An IEEE Computer Society standards committee since2005.
Has specifications in the areas of;
agent applicationsabstract architectureagent communicationagent managementagent message transport
![Page 96: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/96.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
FIPA ACL
FIPA ACL (Agent Communication Language): aspecification for inter-agent communication via messagepassing.
Assumes that agents have Beliefs, Desires and Intentions(BDI model)FIPA ACL is similar in design to KQML, but gives anexplicit BDI-based semantics to the performatives.
Semantics of inform
“S informs R that P” requires that:
S believes that some proposition P is true,
S intends that R also comes to believe that P is true, and
S does not already believe that R has any knowledge of thethe truth of P.
![Page 97: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/97.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
FIPA ACL
FIPA ACL (Agent Communication Language): aspecification for inter-agent communication via messagepassing.
Assumes that agents have Beliefs, Desires and Intentions(BDI model)
FIPA ACL is similar in design to KQML, but gives anexplicit BDI-based semantics to the performatives.
Semantics of inform
“S informs R that P” requires that:
S believes that some proposition P is true,
S intends that R also comes to believe that P is true, and
S does not already believe that R has any knowledge of thethe truth of P.
![Page 98: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/98.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
FIPA ACL
FIPA ACL (Agent Communication Language): aspecification for inter-agent communication via messagepassing.
Assumes that agents have Beliefs, Desires and Intentions(BDI model)FIPA ACL is similar in design to KQML, but gives anexplicit BDI-based semantics to the performatives.
Semantics of inform
“S informs R that P” requires that:
S believes that some proposition P is true,
S intends that R also comes to believe that P is true, and
S does not already believe that R has any knowledge of thethe truth of P.
![Page 99: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/99.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
FIPA ACL
FIPA ACL (Agent Communication Language): aspecification for inter-agent communication via messagepassing.
Assumes that agents have Beliefs, Desires and Intentions(BDI model)FIPA ACL is similar in design to KQML, but gives anexplicit BDI-based semantics to the performatives.
Semantics of inform
“S informs R that P” requires that:
S believes that some proposition P is true,
S intends that R also comes to believe that P is true, and
S does not already believe that R has any knowledge of thethe truth of P.
![Page 100: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/100.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
FIPA ACL
FIPA ACL (Agent Communication Language): aspecification for inter-agent communication via messagepassing.
Assumes that agents have Beliefs, Desires and Intentions(BDI model)FIPA ACL is similar in design to KQML, but gives anexplicit BDI-based semantics to the performatives.
Semantics of inform
“S informs R that P” requires that:
S believes that some proposition P is true,
S intends that R also comes to believe that P is true, and
S does not already believe that R has any knowledge of thethe truth of P.
![Page 101: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/101.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
FIPA ACL
FIPA ACL (Agent Communication Language): aspecification for inter-agent communication via messagepassing.
Assumes that agents have Beliefs, Desires and Intentions(BDI model)FIPA ACL is similar in design to KQML, but gives anexplicit BDI-based semantics to the performatives.
Semantics of inform
“S informs R that P” requires that:
S believes that some proposition P is true,
S intends that R also comes to believe that P is true, and
S does not already believe that R has any knowledge of thethe truth of P.
![Page 102: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/102.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
FIPA ACL
FIPA ACL (Agent Communication Language): aspecification for inter-agent communication via messagepassing.
Assumes that agents have Beliefs, Desires and Intentions(BDI model)FIPA ACL is similar in design to KQML, but gives anexplicit BDI-based semantics to the performatives.
Semantics of inform
“S informs R that P” requires that:
S believes that some proposition P is true,
S intends that R also comes to believe that P is true, and
S does not already believe that R has any knowledge of thethe truth of P.
![Page 103: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/103.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Protocols in FIPA ACL
Notion of ‘protocol’ refers to stereotyped pattern ofconversation between agents.
Available protocols are usually pre-specified by the agentdesigner.
Agents somehow need to discover which protocol to follow.
Choice of protocol to follow could be negotiated,
but in FIPA ACL, convention is to place name of theprotocol in the :protocol parameter of the message.
![Page 104: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/104.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Protocols in FIPA ACL
Notion of ‘protocol’ refers to stereotyped pattern ofconversation between agents.
Available protocols are usually pre-specified by the agentdesigner.
Agents somehow need to discover which protocol to follow.
Choice of protocol to follow could be negotiated,
but in FIPA ACL, convention is to place name of theprotocol in the :protocol parameter of the message.
![Page 105: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/105.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Protocols in FIPA ACL
Notion of ‘protocol’ refers to stereotyped pattern ofconversation between agents.
Available protocols are usually pre-specified by the agentdesigner.
Agents somehow need to discover which protocol to follow.
Choice of protocol to follow could be negotiated,
but in FIPA ACL, convention is to place name of theprotocol in the :protocol parameter of the message.
![Page 106: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/106.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Protocols in FIPA ACL
Notion of ‘protocol’ refers to stereotyped pattern ofconversation between agents.
Available protocols are usually pre-specified by the agentdesigner.
Agents somehow need to discover which protocol to follow.
Choice of protocol to follow could be negotiated,
but in FIPA ACL, convention is to place name of theprotocol in the :protocol parameter of the message.
![Page 107: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/107.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Protocols in FIPA ACL
Notion of ‘protocol’ refers to stereotyped pattern ofconversation between agents.
Available protocols are usually pre-specified by the agentdesigner.
Agents somehow need to discover which protocol to follow.
Choice of protocol to follow could be negotiated,
but in FIPA ACL, convention is to place name of theprotocol in the :protocol parameter of the message.
![Page 108: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/108.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
FIPA-Query-Protocol
Initiator Participant
query-ref
query-if
agree
refuse
inform-result
failure
inform-t/f[agreed][query-if]
[query-ref]
[agreed]
![Page 109: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/109.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Explanation of FIPA-Query-Protocol
Initiator requests Participant to perform a inform action:
query-if: whether proposition P is true or falsequery-ref: query about specified objects
Participant processes request and decides whether toaccept or refuse.
If decides to accept, “agreed” becomes true.
Participant uses inform-t/f to assert whether P is trueor false;
uses inform-result to refer to object that was queriedabout.
![Page 110: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/110.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Explanation of FIPA-Query-Protocol
Initiator requests Participant to perform a inform action:
query-if: whether proposition P is true or false
query-ref: query about specified objects
Participant processes request and decides whether toaccept or refuse.
If decides to accept, “agreed” becomes true.
Participant uses inform-t/f to assert whether P is trueor false;
uses inform-result to refer to object that was queriedabout.
![Page 111: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/111.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Explanation of FIPA-Query-Protocol
Initiator requests Participant to perform a inform action:
query-if: whether proposition P is true or falsequery-ref: query about specified objects
Participant processes request and decides whether toaccept or refuse.
If decides to accept, “agreed” becomes true.
Participant uses inform-t/f to assert whether P is trueor false;
uses inform-result to refer to object that was queriedabout.
![Page 112: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/112.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Explanation of FIPA-Query-Protocol
Initiator requests Participant to perform a inform action:
query-if: whether proposition P is true or falsequery-ref: query about specified objects
Participant processes request and decides whether toaccept or refuse.
If decides to accept, “agreed” becomes true.
Participant uses inform-t/f to assert whether P is trueor false;
uses inform-result to refer to object that was queriedabout.
![Page 113: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/113.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Explanation of FIPA-Query-Protocol
Initiator requests Participant to perform a inform action:
query-if: whether proposition P is true or falsequery-ref: query about specified objects
Participant processes request and decides whether toaccept or refuse.
If decides to accept, “agreed” becomes true.
Participant uses inform-t/f to assert whether P is trueor false;
uses inform-result to refer to object that was queriedabout.
![Page 114: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/114.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Explanation of FIPA-Query-Protocol
Initiator requests Participant to perform a inform action:
query-if: whether proposition P is true or falsequery-ref: query about specified objects
Participant processes request and decides whether toaccept or refuse.
If decides to accept, “agreed” becomes true.
Participant uses inform-t/f to assert whether P is trueor false;
uses inform-result to refer to object that was queriedabout.
![Page 115: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/115.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Explanation of FIPA-Query-Protocol
Initiator requests Participant to perform a inform action:
query-if: whether proposition P is true or falsequery-ref: query about specified objects
Participant processes request and decides whether toaccept or refuse.
If decides to accept, “agreed” becomes true.
Participant uses inform-t/f to assert whether P is trueor false;
uses inform-result to refer to object that was queriedabout.
![Page 116: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/116.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Protocols as FSMs
Protocols are often implemented as Finite State Machines.
Each state represents a stage in the conversation —sometimes thought of as an information state.
Arcs represent the exchanged messages that allowtransition to a new state.
Valid messages depend on the current state of theconversation — represented by outgoing arcs.
Final states represent completed conversations.
![Page 117: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/117.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Protocols as FSMs
Protocols are often implemented as Finite State Machines.
Each state represents a stage in the conversation —sometimes thought of as an information state.
Arcs represent the exchanged messages that allowtransition to a new state.
Valid messages depend on the current state of theconversation — represented by outgoing arcs.
Final states represent completed conversations.
![Page 118: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/118.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Protocols as FSMs
Protocols are often implemented as Finite State Machines.
Each state represents a stage in the conversation —sometimes thought of as an information state.
Arcs represent the exchanged messages that allowtransition to a new state.
Valid messages depend on the current state of theconversation — represented by outgoing arcs.
Final states represent completed conversations.
![Page 119: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/119.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Protocols as FSMs
Protocols are often implemented as Finite State Machines.
Each state represents a stage in the conversation —sometimes thought of as an information state.
Arcs represent the exchanged messages that allowtransition to a new state.
Valid messages depend on the current state of theconversation — represented by outgoing arcs.
Final states represent completed conversations.
![Page 120: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/120.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Protocols as FSMs
Protocols are often implemented as Finite State Machines.
Each state represents a stage in the conversation —sometimes thought of as an information state.
Arcs represent the exchanged messages that allowtransition to a new state.
Valid messages depend on the current state of theconversation — represented by outgoing arcs.
Final states represent completed conversations.
![Page 121: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/121.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Example FSM for a Conversation
Search Order Pay Logout
Login Logout
Order
![Page 122: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/122.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Agent Roles
Protocols give us a way of specifying a class of legalinteractions between agents.
However, we often want to have higher-level ways ofdescribing agent behaviour.
Key notion: role that is assigned to an agent.
Roles determine rights, duties and opportunities.
The role assumed by an agent limits its possible actions.
Example roles in interaction: seller, buyer, auctioneer
Seller must own goods before submitting them for sale.Buyer may submit bids if credit standing is good.Auctioneer may offer goods and accept bids.
![Page 123: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/123.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Agent Roles
Protocols give us a way of specifying a class of legalinteractions between agents.
However, we often want to have higher-level ways ofdescribing agent behaviour.
Key notion: role that is assigned to an agent.
Roles determine rights, duties and opportunities.
The role assumed by an agent limits its possible actions.
Example roles in interaction: seller, buyer, auctioneer
Seller must own goods before submitting them for sale.Buyer may submit bids if credit standing is good.Auctioneer may offer goods and accept bids.
![Page 124: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/124.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Agent Roles
Protocols give us a way of specifying a class of legalinteractions between agents.
However, we often want to have higher-level ways ofdescribing agent behaviour.
Key notion: role that is assigned to an agent.
Roles determine rights, duties and opportunities.
The role assumed by an agent limits its possible actions.
Example roles in interaction: seller, buyer, auctioneer
Seller must own goods before submitting them for sale.Buyer may submit bids if credit standing is good.Auctioneer may offer goods and accept bids.
![Page 125: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/125.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Agent Roles
Protocols give us a way of specifying a class of legalinteractions between agents.
However, we often want to have higher-level ways ofdescribing agent behaviour.
Key notion: role that is assigned to an agent.
Roles determine rights, duties and opportunities.
The role assumed by an agent limits its possible actions.
Example roles in interaction: seller, buyer, auctioneer
Seller must own goods before submitting them for sale.Buyer may submit bids if credit standing is good.Auctioneer may offer goods and accept bids.
![Page 126: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/126.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Agent Roles
Protocols give us a way of specifying a class of legalinteractions between agents.
However, we often want to have higher-level ways ofdescribing agent behaviour.
Key notion: role that is assigned to an agent.
Roles determine rights, duties and opportunities.
The role assumed by an agent limits its possible actions.
Example roles in interaction: seller, buyer, auctioneer
Seller must own goods before submitting them for sale.Buyer may submit bids if credit standing is good.Auctioneer may offer goods and accept bids.
![Page 127: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/127.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Agent Roles
Protocols give us a way of specifying a class of legalinteractions between agents.
However, we often want to have higher-level ways ofdescribing agent behaviour.
Key notion: role that is assigned to an agent.
Roles determine rights, duties and opportunities.
The role assumed by an agent limits its possible actions.
Example roles in interaction: seller, buyer, auctioneer
Seller must own goods before submitting them for sale.Buyer may submit bids if credit standing is good.Auctioneer may offer goods and accept bids.
![Page 128: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/128.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Agent Roles
Protocols give us a way of specifying a class of legalinteractions between agents.
However, we often want to have higher-level ways ofdescribing agent behaviour.
Key notion: role that is assigned to an agent.
Roles determine rights, duties and opportunities.
The role assumed by an agent limits its possible actions.
Example roles in interaction: seller, buyer, auctioneer
Seller must own goods before submitting them for sale.
Buyer may submit bids if credit standing is good.Auctioneer may offer goods and accept bids.
![Page 129: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/129.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Agent Roles
Protocols give us a way of specifying a class of legalinteractions between agents.
However, we often want to have higher-level ways ofdescribing agent behaviour.
Key notion: role that is assigned to an agent.
Roles determine rights, duties and opportunities.
The role assumed by an agent limits its possible actions.
Example roles in interaction: seller, buyer, auctioneer
Seller must own goods before submitting them for sale.Buyer may submit bids if credit standing is good.
Auctioneer may offer goods and accept bids.
![Page 130: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/130.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Agent Roles
Protocols give us a way of specifying a class of legalinteractions between agents.
However, we often want to have higher-level ways ofdescribing agent behaviour.
Key notion: role that is assigned to an agent.
Roles determine rights, duties and opportunities.
The role assumed by an agent limits its possible actions.
Example roles in interaction: seller, buyer, auctioneer
Seller must own goods before submitting them for sale.Buyer may submit bids if credit standing is good.Auctioneer may offer goods and accept bids.
![Page 131: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/131.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Policies
Policies provide high-level description of acceptablebehaviour (or social norms).
Often used for security.
Conversation policies: do not specify types of messagesthat can be exchanged.Rather, conversation policies place constraints on
who can participate in interaction;under what circumstances;whether sub-conversations can be initiated, etc.Policy language: means of stating what agent can/cannotand should/shouldn’t do.
Policy Example (Phillips & Link)
Agents A and B are discussing arrangements for a party for C .Do they stop discussion when C enters the room? This is amatter of policy.
![Page 132: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/132.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Policies
Policies provide high-level description of acceptablebehaviour (or social norms).
Often used for security.
Conversation policies: do not specify types of messagesthat can be exchanged.Rather, conversation policies place constraints on
who can participate in interaction;under what circumstances;whether sub-conversations can be initiated, etc.Policy language: means of stating what agent can/cannotand should/shouldn’t do.
Policy Example (Phillips & Link)
Agents A and B are discussing arrangements for a party for C .Do they stop discussion when C enters the room? This is amatter of policy.
![Page 133: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/133.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Policies
Policies provide high-level description of acceptablebehaviour (or social norms).
Often used for security.
Conversation policies: do not specify types of messagesthat can be exchanged.
Rather, conversation policies place constraints on
who can participate in interaction;under what circumstances;whether sub-conversations can be initiated, etc.Policy language: means of stating what agent can/cannotand should/shouldn’t do.
Policy Example (Phillips & Link)
Agents A and B are discussing arrangements for a party for C .Do they stop discussion when C enters the room? This is amatter of policy.
![Page 134: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/134.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Policies
Policies provide high-level description of acceptablebehaviour (or social norms).
Often used for security.
Conversation policies: do not specify types of messagesthat can be exchanged.Rather, conversation policies place constraints on
who can participate in interaction;under what circumstances;whether sub-conversations can be initiated, etc.Policy language: means of stating what agent can/cannotand should/shouldn’t do.
Policy Example (Phillips & Link)
Agents A and B are discussing arrangements for a party for C .Do they stop discussion when C enters the room? This is amatter of policy.
![Page 135: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/135.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Policies
Policies provide high-level description of acceptablebehaviour (or social norms).
Often used for security.
Conversation policies: do not specify types of messagesthat can be exchanged.Rather, conversation policies place constraints on
who can participate in interaction;
under what circumstances;whether sub-conversations can be initiated, etc.Policy language: means of stating what agent can/cannotand should/shouldn’t do.
Policy Example (Phillips & Link)
Agents A and B are discussing arrangements for a party for C .Do they stop discussion when C enters the room? This is amatter of policy.
![Page 136: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/136.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Policies
Policies provide high-level description of acceptablebehaviour (or social norms).
Often used for security.
Conversation policies: do not specify types of messagesthat can be exchanged.Rather, conversation policies place constraints on
who can participate in interaction;under what circumstances;
whether sub-conversations can be initiated, etc.Policy language: means of stating what agent can/cannotand should/shouldn’t do.
Policy Example (Phillips & Link)
Agents A and B are discussing arrangements for a party for C .Do they stop discussion when C enters the room? This is amatter of policy.
![Page 137: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/137.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Policies
Policies provide high-level description of acceptablebehaviour (or social norms).
Often used for security.
Conversation policies: do not specify types of messagesthat can be exchanged.Rather, conversation policies place constraints on
who can participate in interaction;under what circumstances;whether sub-conversations can be initiated, etc.
Policy language: means of stating what agent can/cannotand should/shouldn’t do.
Policy Example (Phillips & Link)
Agents A and B are discussing arrangements for a party for C .Do they stop discussion when C enters the room? This is amatter of policy.
![Page 138: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/138.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Policies
Policies provide high-level description of acceptablebehaviour (or social norms).
Often used for security.
Conversation policies: do not specify types of messagesthat can be exchanged.Rather, conversation policies place constraints on
who can participate in interaction;under what circumstances;whether sub-conversations can be initiated, etc.Policy language: means of stating what agent can/cannotand should/shouldn’t do.
Policy Example (Phillips & Link)
Agents A and B are discussing arrangements for a party for C .Do they stop discussion when C enters the room? This is amatter of policy.
![Page 139: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/139.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Policies
Policies provide high-level description of acceptablebehaviour (or social norms).
Often used for security.
Conversation policies: do not specify types of messagesthat can be exchanged.Rather, conversation policies place constraints on
who can participate in interaction;under what circumstances;whether sub-conversations can be initiated, etc.Policy language: means of stating what agent can/cannotand should/shouldn’t do.
Policy Example (Phillips & Link)
Agents A and B are discussing arrangements for a party for C .Do they stop discussion when C enters the room? This is amatter of policy.
![Page 140: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/140.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Electronic Institutions, 1
Agent counterpart of human organizations.
Specifies norms and rules to govern interaction.
Conversation protocols are grouped into scenes.
Agents participate in scenes by virtue of a role — can playdifferent roles in different scenes.
Example scenes (for auction):
admit buyersadmit sellerscarry out auctionsettlement (i.e., paying for goods)
![Page 141: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/141.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Electronic Institutions, 1
Agent counterpart of human organizations.
Specifies norms and rules to govern interaction.
Conversation protocols are grouped into scenes.
Agents participate in scenes by virtue of a role — can playdifferent roles in different scenes.
Example scenes (for auction):
admit buyersadmit sellerscarry out auctionsettlement (i.e., paying for goods)
![Page 142: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/142.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Electronic Institutions, 1
Agent counterpart of human organizations.
Specifies norms and rules to govern interaction.
Conversation protocols are grouped into scenes.
Agents participate in scenes by virtue of a role — can playdifferent roles in different scenes.
Example scenes (for auction):
admit buyersadmit sellerscarry out auctionsettlement (i.e., paying for goods)
![Page 143: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/143.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Electronic Institutions, 1
Agent counterpart of human organizations.
Specifies norms and rules to govern interaction.
Conversation protocols are grouped into scenes.
Agents participate in scenes by virtue of a role — can playdifferent roles in different scenes.
Example scenes (for auction):
admit buyersadmit sellerscarry out auctionsettlement (i.e., paying for goods)
![Page 144: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/144.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Electronic Institutions, 1
Agent counterpart of human organizations.
Specifies norms and rules to govern interaction.
Conversation protocols are grouped into scenes.
Agents participate in scenes by virtue of a role — can playdifferent roles in different scenes.
Example scenes (for auction):
admit buyersadmit sellerscarry out auctionsettlement (i.e., paying for goods)
![Page 145: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/145.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Electronic Institutions, 1
Agent counterpart of human organizations.
Specifies norms and rules to govern interaction.
Conversation protocols are grouped into scenes.
Agents participate in scenes by virtue of a role — can playdifferent roles in different scenes.
Example scenes (for auction):
admit buyers
admit sellerscarry out auctionsettlement (i.e., paying for goods)
![Page 146: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/146.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Electronic Institutions, 1
Agent counterpart of human organizations.
Specifies norms and rules to govern interaction.
Conversation protocols are grouped into scenes.
Agents participate in scenes by virtue of a role — can playdifferent roles in different scenes.
Example scenes (for auction):
admit buyersadmit sellers
carry out auctionsettlement (i.e., paying for goods)
![Page 147: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/147.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Electronic Institutions, 1
Agent counterpart of human organizations.
Specifies norms and rules to govern interaction.
Conversation protocols are grouped into scenes.
Agents participate in scenes by virtue of a role — can playdifferent roles in different scenes.
Example scenes (for auction):
admit buyersadmit sellerscarry out auction
settlement (i.e., paying for goods)
![Page 148: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/148.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Electronic Institutions, 1
Agent counterpart of human organizations.
Specifies norms and rules to govern interaction.
Conversation protocols are grouped into scenes.
Agents participate in scenes by virtue of a role — can playdifferent roles in different scenes.
Example scenes (for auction):
admit buyersadmit sellerscarry out auctionsettlement (i.e., paying for goods)
![Page 149: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/149.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Electronic Institutions, 2
Scenes are connected into a performative structure;
latter governs how agents can move from one scene toanother.
E.g., admit buyer precedes auction, auction precedessettlement
Norms govern transitions between scenes.
E.g., a buyer agent that wins a bid is obliged to pay for thegood.
![Page 150: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/150.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Electronic Institutions, 2
Scenes are connected into a performative structure;
latter governs how agents can move from one scene toanother.
E.g., admit buyer precedes auction, auction precedessettlement
Norms govern transitions between scenes.
E.g., a buyer agent that wins a bid is obliged to pay for thegood.
![Page 151: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/151.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Electronic Institutions, 2
Scenes are connected into a performative structure;
latter governs how agents can move from one scene toanother.
E.g., admit buyer precedes auction, auction precedessettlement
Norms govern transitions between scenes.
E.g., a buyer agent that wins a bid is obliged to pay for thegood.
![Page 152: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/152.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Electronic Institutions, 2
Scenes are connected into a performative structure;
latter governs how agents can move from one scene toanother.
E.g., admit buyer precedes auction, auction precedessettlement
Norms govern transitions between scenes.
E.g., a buyer agent that wins a bid is obliged to pay for thegood.
![Page 153: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/153.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Electronic Institutions, 2
Scenes are connected into a performative structure;
latter governs how agents can move from one scene toanother.
E.g., admit buyer precedes auction, auction precedessettlement
Norms govern transitions between scenes.
E.g., a buyer agent that wins a bid is obliged to pay for thegood.
![Page 154: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/154.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Summary
Agent communication requires sharing at a number oflevels:
language for expressing propositional knowledge,language for representing speech acts (performatives),protocols which determine possible messages and theirsequencing.
Higher level structures govern more abstract aspects ofinteraction.
Policy languages and electronic institutions: two ways ofrepresenting rights and obligations of agents.
![Page 155: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/155.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Summary
Agent communication requires sharing at a number oflevels:
language for expressing propositional knowledge,
language for representing speech acts (performatives),protocols which determine possible messages and theirsequencing.
Higher level structures govern more abstract aspects ofinteraction.
Policy languages and electronic institutions: two ways ofrepresenting rights and obligations of agents.
![Page 156: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/156.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Summary
Agent communication requires sharing at a number oflevels:
language for expressing propositional knowledge,language for representing speech acts (performatives),
protocols which determine possible messages and theirsequencing.
Higher level structures govern more abstract aspects ofinteraction.
Policy languages and electronic institutions: two ways ofrepresenting rights and obligations of agents.
![Page 157: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/157.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Summary
Agent communication requires sharing at a number oflevels:
language for expressing propositional knowledge,language for representing speech acts (performatives),protocols which determine possible messages and theirsequencing.
Higher level structures govern more abstract aspects ofinteraction.
Policy languages and electronic institutions: two ways ofrepresenting rights and obligations of agents.
![Page 158: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/158.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Summary
Agent communication requires sharing at a number oflevels:
language for expressing propositional knowledge,language for representing speech acts (performatives),protocols which determine possible messages and theirsequencing.
Higher level structures govern more abstract aspects ofinteraction.
Policy languages and electronic institutions: two ways ofrepresenting rights and obligations of agents.
![Page 159: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/159.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Summary
Agent communication requires sharing at a number oflevels:
language for expressing propositional knowledge,language for representing speech acts (performatives),protocols which determine possible messages and theirsequencing.
Higher level structures govern more abstract aspects ofinteraction.
Policy languages and electronic institutions: two ways ofrepresenting rights and obligations of agents.
![Page 160: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/160.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Reading
http://www.cs.helsinki.fi/u/kraatika/Courses/Agents/fipa-acl.html — useful summary of FIPA ACL.
Wooldridge, esp Chaps 1, 2, 8.
Passin, Chap 9.
![Page 161: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/161.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Reading
http://www.cs.helsinki.fi/u/kraatika/Courses/Agents/fipa-acl.html — useful summary of FIPA ACL.
Wooldridge, esp Chaps 1, 2, 8.
Passin, Chap 9.
![Page 162: Multi-agent Semantic Web Systems: Agent Communication Overvie · 2007. 2. 23. · Multi-agent Semantic Web Systems: Agent Com-munication Overview Jacques Fleuriot Outline Agents Overview](https://reader035.vdocument.in/reader035/viewer/2022071019/5fd27ce53f1e5f289c23f730/html5/thumbnails/162.jpg)
Multi-agentSemantic Web
Systems:Agent Com-municationOverview
JacquesFleuriot
Outline
AgentsOverview
AgentArchitectures
Middle Agents
NetworkArchitectures
Agent Com-munication
InteractionModels
Summary
Reading
http://www.cs.helsinki.fi/u/kraatika/Courses/Agents/fipa-acl.html — useful summary of FIPA ACL.
Wooldridge, esp Chaps 1, 2, 8.
Passin, Chap 9.