DECISION SUPPORT SYSTEMS A KNOWLEDGE-BASED APPROACH

For today's aspiring managers, a practical grasp of computer-based decision support is essential. Try to imagine an organization in which managers are unable to use computers to aid any of their decisional activities. Contrast this vision with one of an organization whose managers routinely employ computers to get at and process knowledge that has a bearing on the decisions being made. These decision support systems store and process certain kinds of knowledge at much higher speeds than the human mind. In addition to such efficiency advantages, they can also be more effective in certain kinds of knowledge handling because they are not subject to such common human conditions as oversight, forgetfulness, miscalculation, bias, and stress. Failure to appreciate or exploit such decision support possibilities can put managers and their organizations at a major disadvantage.

This chapter has laid the foundation for understanding where decision support systems can fit into a manager's activities. We have seen that decision making pervades managerial activities. Thus, aids to decision making can have far reaching impacts on a manager's efforts to excel. We have seen that, as a decision maker, the manager is very much concerned with handling knowledge. This is exactly where decision support systems can help. They automate various knowledge management tasks. Decision support systems are fundamentally concerned with improving the effectiveness and efficiency of knowledge management activities which occur in the course of decision making.

DECISION SUPPORT SYSTEMS A KNOWLEDGE-BASED APPROACH

This chapter has provided a perspective of decision making that is both wide-ranging and fairly detailed. Experiences as a student of management and later as an active manager will serve to enrich and reinforce this perspective. We have seen that decision making is an activity culminating in the choice of a course of action. It can also be seen as a manufacturing activity that produces a new piece of knowledge committing us to a course of action. This knowledge-based view of decision making will be further developed in Chapter 4.

Decisions are not manufactured in a vacuum. They are made within some setting or context. We can consider contextual differences in terms of such factors as management level, situation maturity, decision concurrency, and organization design. All decisions are not of the same type. They can be classified according to such factors as managerial levels, managerial functions, functional area distinctions, degree of structuredness, and presence of negotiation. An appreciation of decision contexts and types can help us understand what features would be useful to have in a decision support system. The same can be said for an appreciation of decision makers and decision processes. These are explored in the next chapter.

DECISION SUPPORT SYSTEMS A KNOWLEDGE-BASED APPROACH

We have seen that all decision makers are not alike. Some are individuals. Others have multiple participants. Some multi- participant decision makers are teams, others are groups, and yet others are organizations. Differences in decision contexts, types, and makers suggest that there may also be differences in decision support systems.

There are certain common traits that decision-making processes tend to exhibit. They typically involve the phases of intelligence, design, and choice. They are very much concerned with the recognition and subsequent solution of problems. A decision-making episode is a flow of problem solving episodes. Decision-making processes are based on strategies to guide the flow. Some common strategies include optimizing, satisficing, and incrementalism. Decision support systems can help in any of the phases, in problem recognition and solution efforts, and in the implementation of various strategies.

The need for decision support systems stems from the realities of cognitive, economic, and temporal limits. The capacity for having decision support systems stems from ongoing technological advances that puts inexpensive, yet powerful, computers on our desktops, in our vehicles, and even in our pockets. It also stems from conceptual and software advances made by decision support researchers and practitioners. The nature of support that can be furnished by a DSS has progressed rapidly over the past decade and is likely to make continued advances. Because knowledge forms the fabric of decision making, all the various kinds of support that a DSS can provide are essentially exercises in knowledge management. Thus, we now take a closer look at the matter of knowledge.

DECISION SUPPORT SYSTEMS A KNOWLEDGE-BASED APPROACH

The topic of knowledge is often taken for granted in discussions of decision making, problem solving, decision support, and computer systems. It is sometimes vaguely equated with the notion of information, masking important distinctions between information and other types of knowledge. Given that knowledge is the object of decision making and problem solving, it is only reasonable that students interested in computer-based decision support should pay attention to how knowledge "is marshalled and managed in it various forms, both within and related to" decision-making processes [van Lohuizen 1986]. The purpose of this chapter has been to highlight a number of significant knowledge matters.

We began with a characterization of basic knowledge flows related to the manufacture of decisions. Summarized in Figure 4-2, these emphasize the fact that decision making is a knowledge-intensive activity. We then grappled with the question "What is knowledge?" This led to several insights about knowledge. It is embodied in representations that are usable (i.e., processable). There are various states of knowledge that result from acquisition and derivation. Knowledge production involves stocks and flows of knowledge. The flows are concerned with acquiring knowledge and deriving knowledge. Both are typically interspersed in a decision-making process and there can be tradeoffs between acquiring and deriving needed knowledge. The source of knowledge can influence its quality. Validity and utility are two major aspects of knowledge quality.

Finally, we turned to the topic of knowledge management. As a field of study, it is concerned with the representation and processing of knowledge. It is related to investigations in the field known as cognitive science. Here, we focused on knowledge management from two angles. First, there are a number of computerized techniques that have arisen with their own distinctive approaches to representing and processing knowledge. A good grounding in decision making and problem solving principles is essential for appreciating the roles of each technique. Second, we examined six important types of knowledge that decision makers need to manage. In ensuing chapters, we shall see how a DSS can aid in the management of each.

DECISION SUPPORT SYSTEMS A KNOWLEDGE-BASED APPROACH

Our overview of decision support systems began by noting that the purpose of a DSS could be viewed as facilitating one or more of a decision maker's abilities, aiding in one or more of the three decision-making phases, improving the flows of problem solving, assisting decision makers in dealing with unstructured and semistructured decisions, or helping decision makers manage knowledge. We then looked at DSSs in historical perspective. The main traits of DP and MIS forerunners were examined as a foundation for understanding what characteristics a DSS shares with other kinds of computer systems and what characteristics distinguish a DSS from these others. We have also traced the contributions of various technological advances to the rise of decision support systems.

Having a feel for the basic characteristics a user could expect to witness in dealing with a DSS, we have considered many of the possible benefits that could result from DSS usage. The extent to which these benefits are provided by a particular DSS depends not only on specific features of that DSS, but also on the natures of the decision maker and the decision situation. Even though they may offer many benefits, we have seen that DSSs are not panaceas. They do have limits of various kinds.

Finally, we turned to a consideration of possible behaviors that could be expected of DSSs. To guide this exploration, the metaphor of a human decision support system was adopted. Four behavioral categories were identified: accepting requests, making responses, possessing knowledge, and processing. For each supporting behavior of a HDSS, we have identified comparable behaviors for a DSS. Some of these are commonly observed in today's DSSs. Others are more rare or only partially exist in DSSs. Further technological advances will determine the extent to which DSSs can more fully provide the services of HDSSs.

DECISION SUPPORT SYSTEMS A KNOWLEDGE-BASED APPROACH

This chapter has introduced the generic DSS framework. From the perspective of this framework, a decision support system can be studied in terms of four interrelated elements: a language system, a presentation system, a knowledge system, and a problem processing system. The first three of these are systems of representation: the set of all requests a user can make, the set of all responses the DSS can present, and the knowledge representations presently stored in the DSS. The problem processor is a dynamic system that can accept any request in the LS and react with a corresponding response from the PS. Which response corresponds to which request is determined by the PPS, often in light of the knowledge available to it in the KS. That is, a change in the KS could very well yield a different response for the same request. As in the case of a HDSS, some DSSs can even produce responses without having received a corresponding request. In addition to reacting to users, they take initiative in the processing of knowledge.

There are many special cases of the generic DSS framework, each characterizing a distinct class of decision support systems. Several of these more specialized frameworks have been examined here. They differ in terms of their emphasis on one or another popular knowledge management technique.

Our survey of specialized frameworks serves several purposes. First, it reinforces an understanding of the generic framework, by illustrating what is meant by a KS, PPS, LS, and PS. Second, it offers an overview of important kinds of DSSs. Third, the survey gives a brief introduction to various knowledge management techniques: text management, database management, spreadsheet management, solver management, and rule management. Each of these, plus various techniques for managing presentation and linguistic knowledge, receives additional coverage in Chapter 9. Fourth, it provides a helpful background for the next chapter's discussion of building decision support systems.

DECISION SUPPORT SYSTEMS A KNOWLEDGE-BASED APPROACH

This chapter has explored issues concerned with the building of decision support systems. In particular, we have looked at who it is that builds DSSs and what steps might reasonably be taken to transform a recognized need for a DSS into an operational DSS. The spectrum of DSS developers ranges from novice do-it-yourself developers to experienced professional developers. In either case, successful DSS development depends on understanding what the end user needs, familiarity with the problem domain, knowledge source access, appropriate tool selection, and skill in using computer technology. The latter two are typically strengths of a professional developer, while the others tend to be the strongest points of a do-it-yourself developer. The technical skills of do-it-yourself developers can be strengthened by books such as this and supplemented by technical assistance from information centers.

The process of do-it-yourself development begins with the recognition of a need or opportunity for decision support. This leads to a setting of broad objectives and evaluation standards for the envisioned DSS, plus a plan for the DSS development project. A typical development project includes the phases of analysis, design, and implementation. Analysis involves the production of a detailed set of requirements that the prospective DSS should meet. These include functional, interface, and coordination requirements.

The design phase is an exercise in figuring out how the system requirements can be satisfied, usually with respect to the facilities furnished by a selected development tool(s). One way to structure the design effort is in terms of DSS architectural elements: LS design, PS design, KS design, and PPS design. Often, because of the tool used, there is no PPS design needed and LS or PS design tends to be minimal. The bulk of design effort tends to be concentrated on coming up with a blueprint for the KS.

In the implementation phase, a developer uses a selected tool(s) to transform DSS designs into an operational system. This usually concentrates on depositing knowledge into the KS according to the KS blueprints. Before the DSS is operational, a developer tests it to check whether it behaves as expected. If it does not, corrective action is taken. Implementation also entails documentation of the DSS to aid in smooth operation. An implemented DSS becomes operational when it is installed for use in a work environment. During the operational phase of its life, a DSS can undergo incremental modifications and redevelopment. It is also subject to general administration.

For a do-it-yourself developer, some or most parts of the development process outlined here may be carried out more or less informally. In the next chapter, we delve into the natures of development tools in more detail, classifying them by knowledge management techniques, roles in development, interface styles, and integration traits provided.

DECISION SUPPORT SYSTEMS A KNOWLEDGE-BASED APPROACH

Development tools are essential for building DSSs. The tool(s) chosen for developing a particular DSS strongly influences not only the development process, but also the features that the resultant DSS can offer to a user. Available tools can be examined from several vantage points that help in understanding their influences on the process and product of DSS development. This chapter has presented four of these perspectives: technique orientation, role in development, interface styles allowed, and integration approaches.

A particular tool is oriented toward one of more knowledge management techniques. Conversely, a particular technique (in its many possible variants) is offered by more than one development tool. Thus, tools can be categorized in terms of the knowledge management techniques they furnish. A spreadsheet tool offers some variant of the spreadsheet technique for knowledge management, a database tool provides some variant of a database technique for managing knowledge, and so on. Although many tools tend to emphasize one technique or another, vestiges of additional techniques are often apparent. Some tools furnish healthy doses of multiple techniques (e.g., as with a spreadsheet tool that is also a database tool).

A different way to classify tools is based on their roles in a development process. We can distinguish among 1) an intrinsic tool which will serve as the PPS of the developed DSS, 2) a partially intrinsic tool, which will serve a part of the DSS's problem processor, and 3) an extrinsic tool, which does not participate in that PPS. An extrinsic tool helps the developer produce all or part of the PPS or to create some portion of the KS contents. Do-it-yourself development trends to rely primarily on intrinsic tools. Tools in the other two categories are primarily of interest to experienced or professional developers.

Development tools can be differentiated with respect to interface styles they allow to be incorporated in DSSs. A DSS's interface is defined in terms of its LS, its PS, its PPS facilities for interpreting, assisting, and packaging, and its linguistic or presentation knowledge held in the KS. With respect to the LS, an interface style refers to the means available for user to make requests. Possibilities include command-oriented, natural language, menu-driven, form-oriented, question/answer, various kinds of direct manipulation, and combinations of these. With respect to the PS, assistance messages can follow the foregoing styles. Results messages can be classified as textual (free-form and structured) versus graphical. Developers are well advised to be familiar with all of these possibilities and to pay close attention to which of them a tool allows.

Another important angle from which to study development tools involves the types of integration they permit within DSSs. This is relevant whenever multiple knowledge management techniques are employed within the bounds of a single DSS. These techniques may be integrated within a single tool or across multiple tools. In the former case, nested and synergistic integration are distinct possibilities. In the latter case, integration can be via a direct format conversion, clipboard, or confederation approach. These five integration styles have various advantages and disadvantages relative to each other. It is not unusual for more than one of them to be employed in a single DSS. Subsequent chapters explore individual knowledge management techniques that are subject to integration.

DECISION SUPPORT SYSTEMS A KNOWLEDGE-BASED APPROACH

In this chapter we have surveyed four knowledge management techniques that can be usefully employed in decision support systems: expression management, text management, hypertext management, and database management. Problem processors that implement expression management are valuable for helping decision makers with ad hoc calculations that arise in the course of decision making. The main objects of interest with this technique are expressions, variables, functions, and macros. These can also be manipulated within the scope of other knowledge management techniques.

A problem processor that implements text management gives decision makers the ability to work with electronic documents. These are not restricted to representing any particular kind of knowledge. However, such representations are processed simply as documents, without concern for the type of knowledge held. In cases where knowledge represented in one document is logically related to that in others, a PPS that implements hypertext management is valuable. This allows knowledge to be organized into an interconnected network of documents called a hyperdocument. The decision maker can follow markers or a map to navigate through the network to access those documents that seem relevant to the decision at hand.

Database management gives a comparatively structured way for organizing knowledge and has historically been used primarily for descriptive knowledge. When a PPS implements database management, it will contain software known as a database control system and the KS will hold one or more databases. Using the relational approach to database management, each database is composed of one or more tables. Each table has a structure defined in terms of fields and a content organized into records. With a query facility, decision makers can extract desired data from a database on the spur of the moment.