Data Exploration or What have those agents ever done for us

1
Data Explorationor What have those agents ever
done for us?

• Alasdair Allan
• University of Exeter, Exeter, U.K.

2
Agent architectures?

• An agent is just software not magic
• Most successful agent architectures are based on
  partial plan approaches
• Most multi-agent systems are based on the
  collaborative agents paradigm

3
Multi-agent systems

• A multi-agent system is one that consists of a
  number of agents, which interact with one-another
• In the most general case, agents will be acting
  on behalf of users with different goals and
  motivations
• To successfully interact, they will require the
  ability to cooperate, coordinate, and negotiate
  with each other, much the same as real people

4
Multi-agent systems are about

• Decision making

• Asking questions

• Brokering agreements
• Carrying out resource discovery could mean that
  your agent looks to your collaborators agent for
  data and expertise before it looks to central
  sources.

5
Data mining

• Collaborative agents operate in a flat world,
  they can access proprietary data from a
  collaborator just as easily as public data from a
  central source

• Agents can encapsulate knowledge, and retrieve
  data that will be good enough to get the job
  done

6
Data mining

• Encapsulating knowledge in the VO means that we
  can generate high level science products

• High level products can be used as the basis for
  decision making

• Every decision made means that the learning curve
  for the user is that much shallower

7
Data exploration

• Astronomers will want to combine their own data
  with data from the VO

• Proprietary data, both their own, and data
  belonging to collaborators

• New data and event notification in real time
  directly from the observatories

8
What they have done for us

• Agent architectures proved to be a big win for
  robotic telescopes networks

• Using agent architectures increased our
  flexibility, scalability and autonomy

• Real time systems needed robustness, agent
  architectures provided it

9
Suggestions?

• I think we should investigate science and
  infrastructure use cases for agents

• Build prototypes for resource discovery and data
  exploration to see if they can provide a similar
  win for the VO