Newsgroups: comp.ai.shells,comp.ai
Path: cantaloupe.srv.cs.cmu.edu!nntp.club.cc.cmu.edu!goldenapple.srv.cs.cmu.edu!das-news2.harvard.edu!news.dfci.harvard.edu!camelot.ccs.neu.edu!news.mathworks.com!rill.news.pipex.net!pipex!netcom.net.uk!nntpfeed.doc.ic.ac.uk!sunsite.doc.ic.ac.uk!nntp0.brunel.ac.uk!strath-cs!dcl-cs!usenet
From: Jonny Carlos da Silva <jonny@comp.lancs.ac.uk>
Subject: Re: Knowlegde / Rules organisation
Content-Type: text/plain; charset=us-ascii
To: nykjz@ny.ubs.com
Message-ID: <33392D41.873@comp.lancs.ac.uk>
Sender: usenet@comp.lancs.ac.uk
Nntp-Posting-Host: engc45.comp.lancs.ac.uk
Content-Transfer-Encoding: 7bit
Cc: Peter.Kohut@ny.ubs.com
Organization: EDC- LANCASTER UNIVERSITY-UK
References: <332E94E5.76E8@ny.ubs.com>
Mime-Version: 1.0
Date: Wed, 26 Mar 1997 14:05:53 GMT
X-Mailer: Mozilla 3.0 (Win95; I)
Lines: 216
Xref: glinda.oz.cs.cmu.edu comp.ai.shells:3620 comp.ai:44991

Peter V. Kohut wrote:
> 
> I'm thinking of developing a "large scale" expert system with CLIPS and
> other tools. Since I have no experience in such large scale knowledge
> projects I was wondering if somebody could give me some hints or give me
> some pointers to books/papers/resources for the following problem :
> 
> Let's assume you want to build a system with approximately 100'000++
> rules. How do you organise these rules to keep the maintenance as low as
> possible ? Are there specialised "knowledge database" tools around? How
> suitable is an object-oriented approach ?
> 
> Any help is appreciated.
> 
> -Peter-

Hi Peter,

Presently, I am working at the Engineering Design Centre- Lancaster
University, as a Visiting Scholar from the Federal University of Santa
Catarina, Brazil, in which I am also lecturer at Mech. Eng. Department. 

This work corresponds to my doctorate degree thesis, which consists of
the development of an Expert System for Fluid Power Systems Design
focusing in the Concurrent Engineering Aspects.

The system is being developed in CLIPS, it has several modules and is
running in Windows (CLIPS 6.0, standard interface) and UNIX (through a
MetaCard Interface). The system has been developed using COOL (CLIPS
Object-Oriented Language) and so far this approach has proved adequate.

The greatest issues to be decided about the system are pointed in the
summary below.
NOTE: some parts of this summary are GENERAL CONCEPTS about 
expert systems that were necessary to present for A SPECIFIC 
AUDIENCE, in order to establish a relation with the topic FLUID 
POWER. Then, if YOU WANT SKIP THEM...

I have submitted a paper to the International Conference on Engineering
Design (ICED97) about the system. This paper is related to the
development process of the expert system. There is an option to download
it from my personal homepage in Brazil.

If you think this project can be of interest, and/or you want to
interact, please contact me.

Regards,

Jonny

Engineering Desing Centre- Lancaster University- Lancaster LA1 4YR- UK
EDC Homepage: http://www.comp.lancs.ac.uk/edc/edc.html
Personal Homepage in Brazil http://www.grante.ufsc.br/jonny/jonny.html

**********************************************************************
Summary of the project objectives presented to British Fluid Power 
Association (BFPA), Technical Committee, Feb. 04th ,97

Application of Schemebuilder Environment for
Hydraulic Systems Design Focusing on 
Concurrent Engineering Aspects


Introduction
	Schemebuilder is a software tool under development at Lancaster
University aimed at getting better concept designs faster. In this
Concurrent Engineering scenario, individuals from different areas in the
industrial environment (such as, design, manufacturing, suppliers,
quality control, etc.) interact during the product development from the
early stages (Conceptual and Preliminary Design). 
Those stages are characterized by two important aspects. Firstly, as
stated by practitioners of concurrent or simultaneous engineering, the
decisions taken during these early stages have the greatest impact on
the product life-cycle. Secondly, these stages have the highest level of
information abstraction, since the design evolves from the user needs
and requirements to the system specification.
	Therefore, the main purpose of Concurrent Engineering is to shorten the
product development time, including design, keeping a better quality and
avoiding rework in the later stages.
	The above mentioned aspects support the application of computer-based
tools, such as expert system, to facilitate the interaction among the
individuals participating in a product design environment.

What is an Expert System?
An Expert System is a computer system which emulates the decision-making
ability of a human expert. The term emulates means that an expert system
is ideally intended to act in all respects like a human expert.
Together with other areas in Artificial Intelligence (Robotics, Natural
Language Processing, Pattern Recognition, Neural Nets, etc.), expert
systems have already proved to be of great value for research and for
industry, with systems such as MYCIN for medical diagnosis  and
PROSPECTOR for mining exploration achieving successful outcomes among
their respective user communities.
Despite this success, not all areas are suited to expert system
development. Thus, the knowledge domain definition is one of the key
issues in the early stages of the development. The following questions
help to clarify and systematise the process of identifying whether there
are likely to be benefits from undertaking the expert system approach.
-When is it Possible?
There are reliable experts in the chosen domain.
The experts must be able to explain the task to a beginner
(non-specialist).
The task should be well understood. 

-When is it Justifiable?
The solution must be valuable.
The human expert is becoming rare.
The expert knowledge is needed in different locations or in adverse
conditions.
-When is it Appropriate?
Symbolic manipulation through rules from experience can be applied.
Sufficiently 
difficult to compensate the investment;
restricted to be manageable;
comprehensive to embrace a practical interest.

What would an Expert System look like in this field?
In the design area, an expert system should provide, among other
aspects:
-Inputs definition according to user needs, including several features,
such as functional, economical and environmental aspects.
-Capacity to generate and evaluate alternative solutions to the same
problem. Establishing a methodological framework for the design process.
-Links to other computer based systems, such as, simulation and database
applications, providing a wider range of analysis for the designer.
-In the Concurrent Engineering context, an expert system should have an
comprehensive knowledge base, with rules from different areas involved
in the design, such as, cost, maintainability, availability and so on. 

All these aspects are being taken into account in the present
development. This explains why the industrial expertise is one of
paramount issues in this project. Further, the design engineer as end
user of the expert system must contribute to its creation from the
earliest stage. In this way the quality of the design provided by the
system assistance will be enhanced.

What is Schemebuilder?
The Lancaster Engineering Design Centre (EDC) primary focus is a
knowledge-based design environment called Schemebuilder, which is a
comprehensive suite of software tools aimed at supporting the designer
in the rapid development of conceptual product designs.
 
Why are Expert Systems relevant to the Conceptual Design and first stage
simulation of Fluid Power Systems?
			
-The hydraulic systems design area has a well established theoretical
foundation. This aspect is important for a pilot-project focusing on
Concurrent Engineering, as well as for Expert System Development. Both
are closely related to the Schemebuilder Project.

-The hydraulic systems are composed of circuits, each having a specific
function, this facilitates the building of functional blocks and also
the application of Object-Oriented Methodology (definition of
Classes-&-Objects, Attributes and Methods).

-Hydraulics is a very broad area covering many fields, including mobile,
machine tools, marine and avionics. Therefore this application will show
the Schemebuilder potential without restricting it.

What is needed to build this Expert System (information, how it is
obtained, how generalised) ?

Due to the complexity involved in developing an Expert System, it is
necessary to establish a very specific knowledge scope application. For
example in the present context, this means to define a relevant field in
hydraulics, e.g.: fork lift machines or excavators, according to the
closest (the nearest and the most interested) companies to be involved
in this process.
In this project, the Concurrent Engineering Focus of which Schemebuilder
is a part must demonstrate attractive tangible benefits to gain the
interest of the companies. This project aims to establish the product
functional structure as well as the design process flow.

Background Questions to Industrial Experts

-What are the time scales in introducing new products, particularly at
the design concept stage and how could they be shortened?
-What use is made of formal techniques such as DFM, FMEA, QFD and how
relevant/beneficial are they?
-In general terms, how is the cost structure of a product determined at
early stages of design?
-How and at what stage of design is maintainability considered?

Assessment of Potential Interest
-As an aid to implementing Concurrent Engineering and shortening
time-to-market.
-As a rapid means of raising and assessing design alternatives.
-As a means of assisting customers with their decision making processes,
by making appropriate parts of the system available to them.
-"Technology Credit" in the market place, by being known to be using
Expert System and Artificial Intelligence Systems.



Some Questions to Industrial Experts that can be useful to implement
this Expert System
-How is the Cost-Structure of the products predicted from the early
stages?
-What maintainability metrics do you use for the products? And in which
stage of the design process they are considered?
-Are you using techniques such as DFM, FMEA, QFD in your current way of
design? If yes, how are they being applied? If not, Why?
-Usually, how long does it take to your company to conclude a typical
design? Is it satisfactory? If not, what is your target, and how do you
plan to achieve it?

What This Tool will be able to offer to Your Company?
-A marketing opportunity to divulge your products through a widespread
database.
-Company profile as linked to a high-tech field, such as Artificial
Intelligence.
-The tool could be offered to your customers to ease their decision
making process. 
-It can be used to start up the implementation of Concurrent Engineering
Aspects in your design process, in case it has not been adopted, or to
increment its application.
