Geoscience Reference
In-Depth Information
supports various representation approaches, can be divided into three main categories (Turban
1995; Durkin 1997; Jackson 1999; Giarratano and Riley 2005; Padhy 2005):
1. ES shells: An ES shell is a special purpose tool for developing an ES in an efficient and
user-friendly environment which provides the basic components of an ES (inference
engine, explanation facility, user interface, etc.), but the knowledge base must be built
by the developer. Shells are the most popular tools for developing ES. There are many
commercial (e.g. EXSYS, XpertRule) and open-source (e.g. OpenRules, Euler) ES shells
available.
2. High-level programming languages: In the context of ES, these are usually symbolic
manipulation programming languages designed for AI applications. Examples of such lan-
guages are Lisp, OPS5 and Prolog.
3. Knowledge engineering tools (or environments): These tools involve a language plus asso-
ciated utility programs to facilitate the development, debugging and delivery of knowl-
edge-based application programs. Popular tools in this category include CLIPS and KEE.
It is accepted that the easiest and most efficient way to develop an ES is to use an ES shell followed
by the knowledge engineering tools and then an AI language. Although these specialised ES shells
may be efficient for developing stand-alone applications, there are limitations in terms of flexibility
and communication when developing hybrid systems, for example, combining ES with GIS for
the development of spatial ES. As a result, the use of a conventional programming language for
developing a hybrid system may provide greater development flexibility even though it is a time-
consuming task (Lukasheh et al. 2001). It is worthwhile to note that a new application of ES for
automated computer program generation has recently been commercially introduced. Funded by
a US Air Force grant, the ES-based application called hprcARCHITECT is capable of generating
computer programs for mixed processor technology systems without the need for technical special-
ists (Feldman 2011).
11.5.3 S
ySteM
e
Valuation
This stage involves system verification and validation using a case study. Verification is defined
by Adrion et al. (1982) as the demonstration of the consistency, completeness and correctness of
the software, that is, building the system right or correctly (O'Keefe et al. 1987). Validation, on
the other hand, is defined by Adrion et al. (1982) as determination of the correctness of the final
software in terms of user needs and requirements, that is, building the right system (O'Keefe et al.
1987). O'Keefe et al. (1987) emphasise that computer professionals concerned with the development
of the system rarely deal with more than verification and validation, the so-called V&V. However,
there are additional system quality issues such as credibility, assessment and evaluation. Credibility
is defined as the extent to which a system is believable or the user can place credence in the system
(Balci 1987). Assessment is the umbrella of issues that considers the
it
between the system and the
user independently of the quality of the decisions made. Evaluation reflects the benefits in terms of
value for money to the users, to the sponsors and generally to the organisation that developed the
system. An empirical evaluation that identifies the characteristics of successful ES is provided by
Metaxiotis et al. (2006).
11.6 RELATIONSHIP BETWEEN ES AND DECISION SUPPORT SYSTEMS
Decision support systems (DSSs) are interactive systems that aid decision-makers in solving semi-
structured and ill-structured decision problems using modelling technology (Gorry and Morton
1971; Turban et al. 2005). As noted by Rodriguez-Bachiller and Glasson (2004), the term DSS
is used frequently in the literature because it has some implicit meaning and is applied to many
