|
The project would focus on multi-site e-Science collaboration and
activity/process/task support. Intelligent (Grid-enabled) agents would
be used in to assist individuals, meetings and group processes and
workflow to support tasks and collaboration intelligently and safely.
This would incorporate research on the following:
- intelligent agent technology
- advanced knowledge technologies, ontlologies, capability descriptions, etc.
- use AI domain, capability and task models
- use methods from the AI planning community
- use methods from the intelligent process management/workflow community
- employ knowledge modelling, semantic web and ontology technologies
- use and provide feedback on emerging standards for the web and grid
such as XML, XML Schema, XSL, RDF, DAML+OIL, UPML
- test (but not develop) different Grid frameworks for this
- test (but not develop) Java/Jini Grid technologies already in use by some
of the participants
- incorporate (but not further develop) collaboration tools,
videoconferencing aids, data walls, etc.
|
|
The project would conduct a series of integrated feasibility demonstrations
(IFDs) which would focus the work with demonstrations on a frequent basis
that integrated more functionality as the research progressed. An initial
demonstration would be promised within 6 months of start to link available
technologies, and 6 monthly thereafter.
There could be links to all 3 of the other IRCs involved in Grid
projects. The project could have demonstrators involving scientific
collaboration between at least AKT and Equator and a life critical
scenario involving Medical Imaging and DIRC.
This research would be tested using two or three (perhaps a fourth later)
applications, with the potential for escalation of each, (depending on
the other IRCs and the associated user/technology organisations
interested in the work):
- Collaboration between partners for intelligent meeting support,
technical exchange, AKT Club dissemination, etc. This could use a
tool base such as the open source Collaborative Virtual Workspace
(CVW) originally from Mitre.
This would be extended
with an "intelligent blackboard" for each participant (possibly via PDAs),
each "room" and each building (site) that uses agent profiles to
help direct interaction and collaboration (again, in collaboration with
Equator).
- Sample scientific research collaboration scenario
Support to colaboration and knowledge sharing between scientific
research groups e.g., for technical meetings and teams via
intelligent agents and intelligent meeting rooms (as in I-Room
proposal under AKT)
- Medical protocol (procedure) support for a task requiring
emergency medical
image interpretation and a set of best practice protocols (in
collaboration with the Medical Informatics and DIRC IRC)
- Sample life critical scenario
mountaineer problem in remote area
personal (PDA) support
local emergency procedural support
escalate to use distributed specialised medical teams
- Emergency response while monitoring care in the community situations.
- The context would be an advanced monitoring system for elderly or
at risk patients wo are being looked after in their homes or
communities rather than hospitalised. Apprpriate local, regional and
national resoitrces could be called upon to repond in a timely and
appropriate way to a range of emergencies. These would account for
the contxt of the individual involved and the resources available to
hand and elsewhere.
- Sample care in the community scenario
Ideas for developing this scenario to include an information
gathering and interpretation element from Tom Rodden includes the
following:
- Emergency Services and support for emergency services where there are
devices in the field that measure biometric material need to access a wide
range of records (including existing medical records and other incident
records).
- Home based support for Well being. Where a range of devices are provided
to support longitudinal recording with a view toward preventative medicine.
This would work toward some sense of real time epidemiology where trends can
be collected remotely etc..
- Medical devices for developing nations. Malcolm Atkinson at Glasgow was
saying he had existing support to help provide medical support where low
cost devices would be placed in developing nations allowing remote monitor,
diagnosis and expertise. The aim would be to maximise access from sets of
central services.
- There is a possibility of another larger scale scenario to link
into the proposed "RoboCup Rescue" challenge scenario for disaster
relief in events such as the Kobe earthquake in Japan. A simulator is
being built for such a scenario (by USC/ISI) that will allow command,
planning, control and communications systems to be tested and evaluated
(as well as robotic aids). A possible link to research on what Austin
Tate calls "showflakes" - massively deployed simple sensors and
on-the-ground agent systems which can self configure and provide
information and local knowledge (via a computing Grid) to higher
knowledge-level task and planning aids is a possible development of
this work in cooperation with other international partners and those
interested in information fusion.
- Sample disaster relief scenario
RoboCup Rescue
International Interdiscplinary Cooperative Research
and challenge problem (see
diagram)
We envisage that AKT could usefully collaborate with other IRCs in the
following manner:
- Links to Equator for collaboration, multi-media, agents: Using CVW
(or other relevant open source based collaboration tools) to link
multiple sites (buildings) with multiple specialised functions (rooms)
using specialised tools and capabilities (agents) would utilise both
the agent-based capabilities of EQUATOR and the infrastructure and
planning technologies available through AKT.
- Links to Medical Imaging for specialised Grid data
interpretation over the Grid in time and life critical situations.
- Links to DIRC for dependability of Grid agents. Users of the grid
will require some notion of agency for tasks such as automated data
mining; information gathering; and decision support based on
distributed knowledge sources. We want these sorts of agent system to
be dependable in various ways but "dependability" in highly
autonomous systems like these is not the same as traditional
dependability in high integrity software systems. Agent systems are
designed to be connected together by loose associations and are built
to be tolerant to faults and misbehaviours of other agents. It is
known to be very difficult to predict the emergent behaviour of such
systems in general so it is important to know if some ways of building
them lead to greater dependability (measured statistically, through
proof, via trusted architectures or by empirical experiment). In
safety-critical systems (such as the second of the application
scenarios above) there is the additional requirement of a safety case,
involving arguments of dependability, that the behaviour of such
systems is controllable within acceptably safe limits. we do not yet
know how to make a safety case for this type of system.
Links
Possible Team
Resources possible are around 3-4 RAs over 3 years starting
October 2001?
Suggestion....
- AKT Edinburgh PIs could John Kingston (medical application),
Austin Tate (both applications),
John Levine (emergency assistance),
Dave Roberston (as AKT link to DIRC)
- Southampton PIs could be Nigel Shadbolt
- Open University - Enrico Motta (reusable and configurable reasoning
services) and Simon Buckingham-Shum (Collaboration - Compedia)
- Equator - Southampton - Dave DeRoure
- Medical Imaging - PI could be be Mike Brady (Oxford)?
- DIRC - PI could be Cliff Jones (Newcastle)?
Possible User & Technology Organisation Interest
- DERA Malvern - Mike Kirton - interest in
agent technology and applications.
- Intelligent Applications Ltd, Livingstone, Scotland - Rob Milne -
interest in distributed task support and mountaineering rescue
scenarios. Rob was involved in mountain rescue in Colorado.
- Faramarz Farhoodi, Tampa, Florida - interest in agents for
the coordination of distributed decision making.
Possible Project Structure
Overall Name: AKTinG (Advanced Knowledge Technologies in the Grid)
- Monologue- Agent Knowledge Technologies in the Grid (AKTinG)
AKT Only
Focus: Knowledge Agents on the Grid
Application in current AKT challenge problems and test cases
- Dialogue - provision of AKT in collaborative environments (prAKTice)
AKT + Equator
Focus: advanced knowledge based collaboration
Application to AKT scientific collaboration and
AKT/Equator collaboration
- Full Cast - Medical prAKTice
AKT + Equator + DIRC + Medical Imaging
Medical Informatics scenarios including use of
collaboration tools for medical consultations and
for medical life critical rescue emergency tasks
- World Stage - Disaster Relief (:-)
AKTinG + International Programmes
Disaster relief joint research and challenge
scanario. Links to coalition research.
Links to RoboCup Rescue.
Notes on Scenarios
Notes provided by Rob Milne, IA Ltd on 22-May-2001...
It is winter. A few hill walkers in the Cairngorms on the Munros (maybe
the peak Beinn Mheadhoin) above Derry Lodge from Linn of
Dee, west of Braemar. The usual storm comes in and they get separated.
wandering in the mist, one person slips down a big snow slope and
breaks a leg and has head injuries. Meanwhile another descends north
and starts to head out that glen, but ends up taking 2 days (that has
happened many times). Another goes over the east side, finds
themselves above a cliff and decides to sit it out (also common). One
of them calls for help in a mobile phone. [ bad weather, tough
approach, but can be reached from 3 directions, it will take a while
to realise that one or two are lost] Lots of searching, plus bad
injuries, time delays because of the bad weather.
Then the storm lifts and they get a chopper and medical team in. Still
searching for the others ...
Then, the helicopter crashes. A couple more serious injuries. Plus
the weather guys say more storms are coming ... could turn into a big
multi-day epic with 100 people and need for remote diagnosis and
medical assistance pretty fast!
Unstructured Collaboration - Compendium
Simon Buckingham-Shum (OU) on 15-Jun-2001 wrote:
I have been working with US colleagues (for the last decade now) on a KM
approach and technology (called Compendium) to support the gathering and
integration of requirements across diverse stakeholders. It was used, for
instance, by Bell Atlantic to underpin their Y2K Contingency Planning:
"What do we do if all the phones in New York State go down on
January 1st?..." [1]
It strikes me that Compendium could find application for bringing
together the many agencies involved in the proposed Disaster Relief
Scenario, to assist communication, and in order for us to define what AKT
could offer. By coincidence, I am already in talks with one relief
agency.
(AKT NOTE: I'm also planning to integrate Compendium into the emerging
suite of technologies. As illustrated in [2-3], it's already
talking to D3E.)
[1] Selvin, A. M., & Buckingham
Shum, S. J. (2001, in press). Rapid Knowledge Construction: A Case Study
in Corporate Contingency Planning Using Collaborative Hypermedia.
Journal of Knowledge and Process Management. PrePrint:
<http://kmi.open.ac.uk/tr/papers/kmi-tr-92.pdf>
[2] Selvin, A., Buckingham Shum, S., Sierhuis, M., Conklin, J.,
Zimmermann, B., Palus , C., Drath, W., Horth, D., Domingue, J., Motta,
E., & Li, G. (2001). Compendium: Making Meetings into Knowledge
Events. Knowledge Technologies 2001, (March 4-7, 2001, Austin
TX).
<http://www.compendiuminstitute.org/compendium/papers/Selvin-KT2001.pdf>
[3] Selvin et al (2001). Participatory, Real Time, Hybrid Knowledge
Capture. Submission to K-CAP 2001.
Page maintained by a.tate@ed.ac.uk,
Last updated: Fri May 10 17:05:16 2002
Medical image via FireViewer
on a Palm Handheld
Mountain rescue at night by helicopter image from
The Mountaineering
Council of Scotland
