DSS DESIGN

As mentioned in Section 2.2.4.1, the DSS development can be affected by the general computer information system development model. The DSS is a computer system that is used to help making decisions, while the general computer information system is a computer system that is not specifically developed for decision making. Software engineering is the establishment and use of sound engineering principles in order to obtain economical software that is reliable and works effectively on real machines (Pressman, 1997), and used in both in general computer information systems and DSSs. Therefore, the general information system design model is introduced first in this section, followed by the theory of software engineering for DSS design. Finally, the DSS design model is reviewed.

(a) Information System Design Model

According to Shelly et al. (2001), there should be three steps for the information system design. These three steps are:

• reviewing system requirements mainly from system planning and system design point of view.

• designing the system in user interface, input processed, input and output formats and reports, data and system architecture.

• presenting the system design.

These steps are based on the SDLC in information system domain. (b) Software Engineering Design Model

Meikle (2002) presented the MODDE (MOdel of Decision Support System Design and Evaluation) methodology for innovations in decision support system design. MODDE uses a

simple idea of the software engineering process to show the way of considering three qualitative extents of high-level independent decision making. These three extents are the preservation of discretion, the maintenance of consistency and the maintenance of resolution (which is the provision of information in an appropriate way) in MODDE, and they have come up directly from a project with the Refugee Review Tribunal (Victoria) in an applied legal domain. MODDE uses a systematic method to address these three extents based on software engineering ideas.

(c) DSS Design Model

From their application of DSS for Forest Management Planning in Canada, MacLean and Porter (1994) had learnt a lot in their DSS design as follows:

• Compatibility with existing (and probable over the next 5-10 years) databases and planning procedures is key to user acceptance and adoption of new DSS tools. In other words, new and better tools will not be used if they are not compatible with the current way of doing business.

• User involvement in the development of DSS tools is critical. The chance of researchers developing something that is close, but not quite right, is extremely high. User involvement continually focuses the attention on real problems.

• A tool-based approach of developing components of an overall system will help by carving a big problem into bite-size chunks, and allows earlier use of and feedback of DSS components. This will ultimately create a "tool-kit" for other DSS developers.

• Consider the management interaction time frame, and plan to have DSS components ready when they can be used.

According to Thierauf (1982), the general steps of DSS design should include: (a) review of new system requirements, (b) incorporate human factors, (c) design the new system and (d) flowchart and document the new system. However, the information below provides more detailed design approaches for DSS design.

Several DSS design approaches were summarised by Yen (1986). These approaches are listed below:

• Carlson’s representation-based approach.

• Courbon's evolution approach.

• Schonberger's contingency approach.

• Sprague and Carlson’s quick-hit approach.

The Carlson’s representation-based approach attempts to reduce the differences between requirements of decision makers and the capacities of the DSS. The Keen's adaptive design approach focuses adaptiveness in a DSS design. The Courbon's evolution approach recommends the progressive design of a DSS by using multiple, minimum-length cycles. The Schonberger's contingency approach offers design leadership. The Sprague and Carlson’s quick-hit approach suggests designing a DSS quickly by taking a recognized high-payoff application area. After studying these methods, Yen (1986) made the first attempt to combine two different research areas, namely, DSS and Expert System (ES), to form a new, combined and improved system design approach (i.e. the use of an Expert System method for DSS design).

Sauter (1997) outlined three approaches to design and implement a DSS. They are (a) one- stage, complete system approach, (b) quick-hit method and (c) evolutionary development method. The latter two methods were described in Yen (1986) as the Sprague and Carlson’s quick-hit approach and the Courbon's evolution approach respectively.

Generally, the one-stage, complete system approach requires building an entire DSS system and then delivering it to the user. There are no subsystems such as models, database, user interface, etc with this approach. This approach is suitable for a large-scale multi-user or unique systems such as a GIS system. A prototype of DSS is often produced with this design method. It is not often used in DSS design today as users want to replace various components with recent developments (e.g. pre-existing tools and models), which is not allowed with this approach.

Both the quick-hit method and the evolutionary development method differ from the one- stage, complete system approach in that the former two methods use the existing technology. The quick-hit method is to design the system quickly for some well-known and useful immediate need. The system is likely to work in a PC environment. The designer can use the existing technologies and other tools, and then concentrate on the analysis and user interface. The evolutionary development method also uses existing technologies, and the designer focuses on the analysis of the needs rather than on the construction of DSS. It differs from

the quick-hit approach as the system design and the system will become better as the user learns from the experience during the system design, development and information access. The immediate feedback from the user is considered in the system design and development. The quick-hit method was used by the author for DSSFCMR in this research project. The author concentrated on analysis and construction of DSS using existing technologies.