A VIEW ON THE WEB ENGINEERING NATURE
OF WEB BASED EXPERT SYSTEMS
Ioannis M. Dokas
Universität Paderborn, Fakultät EIM, Fürstenallee 11, D-33102 Paderborn, Germany
Alexandre Alapetite
Risø National Laboratory, Systems Analysis Department 110, P.O. box 49
Frederiksborgvej 399, DK-4000 Roskilde, Denmark
Keywords: Web based expert systems, Web engineering, Development process model, Web accessibility.
Abstract: The Web has become the ubiquitous platform for distributing infor
mation and computer services. The tough
Web competition, the way people and organizations rely on Web applications, and the increasing user
requirements for better services have raised their complexity. Expert systems can be accessed via the Web,
forming a set of Web applications known as Web based expert systems. This paper supports that the Web
engineering and expert systems principals should be combined when developing Web based expert systems.
A development process model will be presented that illustrates, in brief, how these principals can be
combined. Based on this model, a publicly available Web based expert system called Landfill Operation
Management Advisor (LOMA) was developed. In addition, the results of an accessibility evaluation on
LOMA – the first ever reported on Web based expert systems – will be presented. Based on this evaluation
some thoughts on accessibility guidelines specific to Web based expert systems will be reported.
1 INTRODUCTION
In many real-world problems, expert knowledge is
scarce and yet in high demand. Computer programs
that simulate the thought process of human experts
can propagate domain expertise and can provide a
solution to these problems. These programs are
known as expert systems. In brief, expert systems
have a knowledge base and an inference engine. The
acquired knowledge of a specific domain is stored in
the knowledge base. The inference engine is where
the processing of the stored knowledge and the
reasoning of the expert system take place.
The Internet – and the World Wide Web in
part
icular – has become the ubiquitous platform for
distributing data and information. The rapid growth
of the Web and the evolution of the corresponding
technologies “forced” Web applications to evolve as
complex, challenging and multidimensional projects.
That gave birth to a new engineering approach
known as Web engineering (Deshpande et al., 2002).
Expert systems can be accessed via the Web,
fo
rming a set of Web applications known as Web
based expert systems. Recently, has been mentioned
that Web based expert systems can be considered as
Web engineering projects that can be developed by
merging an expert system and a Web site/application
subprojects
(Dokas, 2005). In essence, Web based
expert systems can be formed by the combination of
at least two components; namely an expert system
and a Web site.
There has been a lack of research and of general
m
ethodology for developing Web based expert
systems (Duan, et al. 2005) and a basic question that
requires an answer is the following: Though which
phases a Web based expert system development
project is progressing, starting from its
conceptualisation and ending when it is released?
This paper is an attempt to answer this question. In
addition, this paper aims to demonstrate the need for
collaborative work among expert systems and Web
engineering practitioners and researchers, on the
development and operation of Web based expert
systems. A development process model will be
280
M. Dokas I. and Alapetite A. (2006).
A VIEW ON THE WEB ENGINEERING NATURE OF WEB BASED EXPERT SYSTEMS.
In Proceedings of the First International Conference on Software and Data Technologies, pages 280-283
DOI: 10.5220/0001310102800283
Copyright
c
SciTePress
presented that reveals the Web engineering nature of
such projects. Based on the presented model, a
publicly available Web based expert system called
Landfill Operation Management Advisor (LOMA)
was developed. Additionally, the results of an
accessibility evaluation on LOMA; the first ever
made on a Web based expert system, will be
presented and some thoughts on accessibility
guidelines specific to Web based expert systems will
be reported.
2 THE PROPOSED
DEVELOPMENT PROCESS
MODEL
Development process models are used for the
description, coordination and management of
necessary activities towards the development,
implementation and release of computer systems.
The proposed development process model is
schematically displayed in Figure 1. It can be used
to help Web based expert system developers to
organize and manage effectively the necessary
activities towards their project completion. Bellow,
each phase of the proposed model will be explained
in detail.
Figure 1: The development process model.
Scope and conceptualization:
Web based expert systems are developed to satisfy a
need that has to do with the delivery of a domain
expertise over the Web. The scope of a Web based
expert system is to satisfy the above-mentioned
need. The general concepts, the necessary
components, the basic working principles and tasks
can be briefly described, forming a conceptual
model.
During LOMA’s conceptualisation phase, the
lack of expertise on landfill operations was realized.
A decision was made to acquire the available
expertise on landfill operational problems and to
propagate it worldwide via a Web based expert
system. High-level restrictions on time and on
expenditures were set. Abstract descriptions of the
system’s components and of the user-system
interactions were made.
Requirements and specifications:
Based on the conceptualization model, a more detail
description of the desired functions of the Web
based expert system should be made leading to a set
of functional requirements. In addition, some desired
characteristics and/or constraints could be revealed.
These are defining the non-functional requirements
of the system. The functional and non-functional
requirements together are defining the Web based
expert system specifications.
In LOMA’s case the initial specifications
included: the overall scope and goal of the Web
based expert system, the target group of users, use
cases describing the Web based expert system
behaviour under various scenarios of user requests,
Risk analysis:
The reason of performing the risk analysis is to
assess the factors that may jeopardize the Web based
expert system development, and to define control
measures and procedures that must be activated
whenever anticipated problems occur during the
development process.
Some of the assessed factors that could
jeopardize LOMA project were the selection of:
improper tools like expert system shells, server
applications, Web editors etc., ineffective
knowledge acquisition techniques, unproductive
development strategy for the Web based expert
system software components. Therefore, additional
specifications were established, which were related
to these issues.
Web Based Expert System Development:
When the preliminary phases reach an acceptable
level of completeness, the development of the Web
based expert systems subcomponents can begin.
These components will include at least an expert
system and a Web site. Based on the specifications,
other computer applications, such as GIS and
database systems, can be developed to enrich the
Web system capabilities. The new challenge behind
Web based expert systems development is to
discover methods and techniques to combine,
coordinate and manage the subcomponents
development in order to save resources and time.
A VIEW ON THE WEB ENGINEERING NATURE OF WEB BASED EXPERT SYSTEMS
281
That is why the subcomponents development phases
are interconnected with two-way arrows in Figure 1.
Each subcomponent development team can
perform the requirements analysis, specifications,
and risk analysis processes again, but this time at a
low level, among other necessary development
processes such as design, coding implementation
and testing. When all subcomponents reach an
acceptable operation level, they can be combined to
form a prototype Web based expert system. Using
the prototype system it is possible to check the
degree to which the specifications are fulfilled.
Web Based Expert System Testing:
Before launching the Web based expert system on
the Web, it is necessary to apply testing procedures
to identify the degree to which the entire Web based
expert system is correct and complete, based on the
specifications. Moreover,
the knowledge base and
the reasoning mechanism of the expert system
component must be tested and validated for errors.
The testing of LOMA was conducted in three
phases. The first was to evaluate to which degree the
specifications have been achieved. The second was
to invite the collaborative domain experts to use
LOMA and to point out problems, bugs, errors or
improvement suggestions. The third was to send e-
mail invitation to landfill experts from different
countries to evaluate LOMA.
Version Release and Evolution:
During this phase, it is important to make sure that
the system can be used properly and some non-
functional requirements can be fulfilled. Since the
domain knowledge, the used tools and the
corresponding technologies are evolving, the Web
based expert system must evolve with time, because
progressively it will be less satisfactory to use.
Recently, the need to improve LOMA Web
based expert system in a way that will make it more
accessible to people with disabilities, expert users
under time pressure and to mobile devises has been
identified. Bellow, the accessibility evaluation of
LOMA will be presented; as far as the authors know,
this is the first ever reported for Web based expert
systems. Through LOMA’s accessibility evaluation
process, some thoughts on accessibility guidelines,
specifically for Web based expert systems, have
been raised and will be presented.
3 WEB ACCESSIBILITY
The Web accessibility concept refers to a combined
set of measures, namely, how easily and how
efficiently different types of users may make use of
a given service. Developers can extend the
portability and longevity of their Web based expert
system by using standard formats and accessibility
rules. Additional improvements are expected, such
as on speed and usability. Accessible Web pages
should indeed be faster to navigate and minimize the
mental load. These factors have in turn an indirect
impact on safety.
3.1 LOMA Accessibility Evaluation
The system has been tested on desktop computer,
running various Web browsers, and on a PDA with a
screen of 230×320px (IE 4.01). Accessibility,
HTML and CSS automatic validation tools were first
used to spot some flaws. Then, the conformance
evaluation of LOMA has been made by following
the W3C WAI methodology and its “Checklist of
Checkpoints for Web Content Accessibility
Guidelines 1.0” [http://www.w3.org/TR/WAI-
WEBCONTENT/full-checklist.html]. On 65
checkpoints, 17 were successful, 27 failed and 21
were not applicable.
Some general flaws were identified in various
categories, such as invalid HTML and CSS code,
character encoding issues, non user-friendly and non
permanent addresses (URLs) and client side
programming (JavaScript) problems.
Accessibility specific issues were for instance
important pictures without alternative text,
insufficient colour contrast, pop-up windows, form
fields without explicit labels, undefined acronyms or
abbreviations, etc.
3.2 Accessibility Considerations for
Web Based Expert Systems
Based on the experience of evaluating of LOMA,
some thoughts on accessibility recommendations
specific to Web based expert systems are reported.
The system should inform the user of the number
of steps already done, and an averaged estimation of
the remaining steps. Furthermore, listing the
questions already dealt with can help users in
establishing a good mental model. Even for expert
users, it is important to have self-explanatory
questions: pictures can greatly improve text
descriptions and are quickly understood. Similarly,
a glossary of the words used in the context of the
current expertise is an interesting resource.
Some Web based expert systems can be used for
time critical decisions, for which there should be a
possibility to go directly to the possible solutions
ICSOFT 2006 - INTERNATIONAL CONFERENCE ON SOFTWARE AND DATA TECHNOLOGIES
282
4 CONCLUSIONS
ranked by probability. Similarly, a mechanism to
mark them as “not applicable” or “unknown” should
be investigated.
In many real life and safety critical circumstances,
there is a need for rapid and effective distribution of
expert knowledge and guidance. Web based expert
systems can be an alternative solution to satisfy that
need. This paper introduced a development process
model for Web based expert systems that shows
the
phases through which a Web based expert system
development project is progressing. In addition, this paper
pointed out that the collaborative work of expert
system, Web engineering practitioners and
researchers is essential in such projects.
In safety critical environments, even more efforts
should be put to ensure robustness. On Image 1, a
list of two advises was designed for large screens
and use a simple carriage return <BR> instead of a
standard list. On a smaller screen, this is rendered as
3 lines, making difficult to isolate the 2 advises.
Using meaningful mark-up helps browsers to ensure
the semantics of the page, even under unexpected
situations. Another robustness issue is due to the
behaviour of some widgets, which are more likely to
show their limitations on small screens. This is for
instance the case with the <SELECT> element in
our tests, which crops the text when it is too long to
fit into the screen, as seen on Image 2. It can be
replaced by a set of radio buttons, or checkboxes.
In an attempt to underline the Web engineering
nature of Web based expert system, an accessibility
evaluation on LOMA has been presented. Based on
LOMA’s accessibility evaluation, some thoughts on
accessibility guidelines specific to Web based expert
systems have been reported. Further work is needed
to validate and test them in different publicly
available Web based expert systems.
Finally, some Web based expert systems use
time limited sessions. Users shall be given sufficient
time, but in safety critical applications, sessions
should be avoided.
ACKNOWLEDGEMENTS
This work is supported by the Fifth European
Framework Programme, within the ADVISES
Research Training Network about “Analysis Design
and Validation of Interactive Safety-critical and
Error-tolerant Systems”. We would like to thank
Henning Boje Andersen, from Risø National
Laboratory, for his comments.
REFERENCES
Image 1: List of items.
Deshpande, Y., Murugesan, S., Ginige, A., Hansen, S.,
Schwabe, D., Gaedke, M., and White, M., B., 2002.
Web engineering. Journal of Web Engineering, 1, 3-
17.
Dokas, I., M., 2005. Developing Web sites for Web based
expert systems: A Web engineering approach. In
Proceedings of the Second International ICSC
Symposium on Information Technologies in
Environmental Engineering (Magdeburg Germany,
September 2005), 202-217, Shaker Verlag.
Duan, Y., Edwards, J., S., and Xu, M., X., 2005. Web-
based expert systems: Benefits and challenges.
Information & Management, 42, 799-811.
Image 2: Content overflow.
A VIEW ON THE WEB ENGINEERING NATURE OF WEB BASED EXPERT SYSTEMS
283