[Search for users]
[Overall Top Noters]
[List of all Conferences]
[Download this site]
Title: | Europe-Swas-Artificial-Intelligence |
|
Moderator: | HERON::BUCHANAN |
|
Created: | Fri Jun 03 1988 |
Last Modified: | Thu Aug 04 1994 |
Last Successful Update: | Fri Jun 06 1997 |
Number of topics: | 442 |
Total number of notes: | 1429 |
259.0. "Anybody need AISE?" by ULYSSE::MCGREGOR () Tue Dec 18 1990 14:54
The AISE System
---------------
As some of you may know, the Valbonne AI group worked a couple of
years ago on a Prolog based-tool which we called AISE. This tool was
used as the basis of the Risk-Advisor and Benchmark-Advisor Systems.
I would like to try to measure interest in this tool, and somehow
"position it" with respect to other tools in the market, to see if it's
worth investing in (other than the minimum to get into ASSETS).
Can you help?
Background:
----------
After letting AISE lie "on the shelf", for a year, we have decided to bring it
down, shake the dust off, and make it available again....
We are actually carrying out the work necessary to put AISE into the
ASSETS library at the end of January 91. This means basically:
- Bringing the documentaion up to date (DW rather than UIS)
- Getting the AISE DW interface to conform to DW "style-guide".
What is AISE?
------------
I do not mention all the features, simply give an idea of the tool!
AISE consists of two components, the knowledge-editor KED, and the
run-time interpreter INF.
KED is a Decwindows editor, which allows you to define objects, and
specify for each object attribute a METHOD which defines how the
attribute derives its value. These methods are written in a language
which can combine the values of other objects logically or
arithmetically. Here is an example method for an object called
BADLY_ORGANISED:
if (PROJECT_MANAGER = no or REQ_SPEC = no) and SIZE^FINANCE > 3 then true
else false
This links the object BADLY_ORGANISED to the objects PROJECT_MANAGER,
REQ_SPEC, and SIZE.
A NETWORK of objects is built up in this way, linked together by the
methods. Methods can also be call-outs, or special methods which ask
the user questions.
You can create special objects called RECOMMENDATIONS which can be
considered to be at the top of the network which is created. At
run-time, these are visible when they are TRUE; there are pieces of text
associated with them, and a component of KED allows you to define how
these are presented to the user at run-time, on-line or in reports.
There are ways in which you can modify the basic control mechanism,
This is very simple, based on propagation of values through the
network.
There are a number of ways you can browse the network in KED to inspect the
knowledge-base.
When you are happy, you "compile" this knowledge, and create a run-time
system.
This gives a simple user-interface allowing modification of values, and
presentation of the status of recommendations, on-line, or in a report
as you defined in KED. The network you created dynamically propagates
value changes up and down the network and maintains consistency. This
is sometimes called a "data-flow" model and you can think of it as
analogous to the way spreadsheets work....
What Types of Problem is AISE Good For?
--------------------------------------
AISE is good for constructing systems which request data in an
"intelligent" way, and summarise or process this data to present it in
a neatly formatted form, and come to some conclusions about it e.g.
- Intelligent Check-list
- A Process Advisor
AISE is intended for incremental development of such knowledge-based
systems.
AISE could be used as a "general purpose" shell, but is not built for
this purpose...
T.R | Title | User | Personal Name | Date | Lines |
---|
259.1 | Test needs | EVTAI1::NGAMBY | L�opold NGAMBY / C.S. Country Support France | Mon Dec 31 1990 13:28 | 4 |
| Befor we can answer to your question about the need of AISE, we need to
reevaluate it in testing his functionalities. So can we have it for
test?
Regards.
|
259.2 | Watch this Space
| GRAPHS::MCGREGOR | | Wed Jan 02 1991 11:55 | 5 |
|
We will make the kit available as soon as the doc is finished, that is to say
at the end of January.
George.
|
259.3 | Still PROLOG-based? | FASDER::MTURNER | Mark Turner * DTN 425-3730 * MEL4 | Tue Jan 22 1991 16:48 | 5 |
| Is AISE still based on PROLOG? If so, do your plans include a
possible move to "C" or other conventional base?
Mark
|
259.4 | C, the universal language?
| TATIA::MCGREGOR | | Wed Jan 23 1991 13:26 | 15 |
| Hi Mark,
The answer is Yes and No: AISE uses Prolog II+, and we have no plans to
port it to C in the short-term. Two points:
1. We do not want to re-implement the features of Prolog in C, which is
effectively what would be needed. My opinion is that people have to get used to
tools and applications built with high-level (AI) languages, (as long as there
are no performance limitations or space problems caused by the high level
language itself).
Otherwise we are only in the "prototype engineering" business...
What do you think?
2. Even if we wanted to do such a port, we don't have the resources!
|
259.5 | to my taste.... | YIPPEE::FITZGIBBON | Joe Fitzgibbon VB/EIC | Thu Jan 24 1991 13:29 | 6 |
|
Ref :-1
I'll raise my glass to thoss sentiments!
Joe.
|