Re: Prolog, expert system shells and inference engines
- From: "girish" <Girish.Padmalayam@xxxxxxxxx>
- Date: 20 Feb 2007 00:53:17 -0800
On Feb 19, 6:07 am, "Kobayashi" <kubilayi...@xxxxxxxxx> wrote:
Hi,
I want to develop expert systems (ES) for medical diagnosis. I have
actually made one by using RealBasic. But as a newbie I still have
some confusions:
1. Do I have to use an "inference engine"? Can not I simply put
IF...THENs? I have built my abovementioned ES in this way. Well, I
must say that it was not easy. But if I used an ES shell would it be
easier?
2. What is the difference between an ES shell and Prolog? Oh, yes, one
is a "shell" and the other is a "programming language". What I try to
mean is which one is better to build an ES?
3. The main question: As far as I know, Prolog has an inference
engine. Right? So it is comparable with ES shells. But today I have
seen a web site in which an ES shell was built by using Prolog:
http://www.cee.hw.ac.uk/~alison/ai3notes/section2_5_4.html
If Prolog has already an inference engine why is it necessary to build
a shell by using it?
Thanks in advance.
There are some characteristics of Prolog's inference engine that
makes one want more control.
The rules are executed top-down and by default it is backward-chained.
This might be alright for some domains but not for others. I am also
experimenting with Prolog after spending some years in Java.
My experiments have started with writing simple facts and rules in
Prolog and checking if the answers match my intuitions. Sometimes the
rules require reordering. You can infer from that you will need finer
control over the inference engine. Naive ordering using a numbering
system is possible. For some people and domains forward-chaining is
more intuitive. That requires writing meta-predicates. Anyway remember
to have fun.
.
- References:
- Prolog, expert system shells and inference engines
- From: Kobayashi
- Prolog, expert system shells and inference engines
- Prev by Date: Call for papers - Workshop on Termination (WST 2007) - Paris
- Next by Date: OOPS Expert System Shell with LEX conflict resolution
- Previous by thread: Re: Prolog, expert system shells and inference engines
- Next by thread: RCRA 2007 - Call for Papers
- Index(es):
Relevant Pages
|
|