ChapterPDF Available

Abstract

This chapter presents a generic architecture that provides terminology for discussing decision support systems and furnishes a unifying framework for guiding explorations of the multitude of issues related to designing, using, and evaluating these systems. The architecture is comprised of four main subsystems: language system, presentation system, knowledge system, and problem-processing system. By varying the makeup of these four elements, different types of decision support systems are produced. Several of the most prominent types of decision support systems are described from an architectural viewpoint.
CHAPTER 9
DSS Architecture and Types
Clyde W. Holsapple
School of Management, Gatton College of Business and Economics, University of Kentucky,
Lexington, KY, USA
This chapter presents a generic architecture that provides terminology for discussing deci-
sion support systems and furnishes a unifying framework for guiding explorations of the
multitude of issues related to designing, using, and evaluating these systems. The architec-
ture is comprised of four main subsystems: language system, presentation system, know-
ledge system, and problem-processing system. By varying the makeup of these four ele-
ments, different types of decision support systems are produced. Several of the most
prominent types of decision support systems are described from an architectural viewpoint.
Keywords: Architecture; Decision support system; DSS; Framework; Knowledge system;
Language system; Presentation system; Problem-processing system
1 Introduction
As the prior chapters suggest, decision support systems are defined in terms of the
roles they play in decision processes. They provide knowledge and/or knowledge-
processing capability that is instrumental in making decisions or making sense of
decision situations. They enhance the processes and/or outcomes of decision mak-
ing. A decision support system (DSS) relaxes cognitive, temporal, spatial and/or
economic limits on the decision maker. The support furnished by the system al-
lows a decision episode to unfold
in more-productive ways (e.
g., faster, less expensively, with less effort),
with greater agility (e.
g., alertness to the unexpected, higher ability to
respond),
innovatively (e.
g., with greater insight, creativity, novelty, surprise),
reputably (e.
g., with higher accuracy, ethics, quality, trust), and/or
with higher satisfaction by decisional stakeholders (e.
g., decision par-
ticipants, decision sponsors, decision consumers, decision implementers)
versus what would be achieved if no computer-based decision support were used.
These concepts are illustrated in Figure 1.
The black box, which represents a decision process, can be thought of as: be-
ing sliced into Simon’s three stages of intelligence, design, and choice; containing
164 Clyde W. Holsapple
a particular decision mechanism such as optimization, elimination-by-aspects, or
nominal group technique; being improvised or having a predefined infrastructure;
being simple and fixed or as a complex, adaptive process; and so forth. As the
two windows into the decision process indicate, the process can involve the ac-
tions of a DSS as well as other participants. The decision sponsor, participant(s),
implementer, and consumer may be distinct individuals; or, an individual may
play more than one of these roles. When a DSS (or multiple DSSs) is involved in
a decision process, it affects the process and its outcome in at least one of the
indicated PAIRS (productivity, agility, innovation, reputation, satisfaction) direc-
tions (Hartono and Holsapple 2004).
Within the foregoing notion of what DSSs are, there is wide variation in terms
of possible DSS application domains, particular characteristics of DSSs, function-
alities designed into these systems, approaches that are offered for interacting with
them, ways in which DSSs are incorporated into decision processes, and kinds of
benefits that accrue from DSS usage. Such variations are examined at length in the
many chapters that follow. This chapter introduces an architecture that is shared
by all DSSs, giving a unified way of thinking about them. Care must be taken to
understand that the architecture does not define what a DSS is; rather, it functions
as an ontology that gives a common language for design, discussion, and evalua-
tion of DSSs, regardless of their manifold variations.
An architecture is essentially a framework for organizing our thoughts about
something. It identifies the major elements to be considered in developing and
using something. The general architecture of houses identifies such important
elements as a plumbing system, an electrical system, an air-treatment system, and
a system of rooms. It also identifies relationships among these elements. Similarly,
the architecture of decision support systems can be described by a generic frame-
work that identifies essential elements of a DSS and their interrelationships. These
elements are various kinds of systems that are configured in a certain way.
Figure 1. The role of a decision support system in decision making
DSS Architecture and Types 165
Here, we begin with an overview of the four generic systems that are basic
elements of any DSS. Their relationships to each other and to the DSS’s users are
shown to be simple and straightforward. We then examine several more-special-
ized DSS frameworks that are special cases of the generic framework. Each char-
acterizes one category of DSSs, such as text-oriented DSSs, database-oriented
DSSs, spreadsheet-oriented DSSs, solver-oriented DSSs, rule-oriented DSSs, and
compound DSSs.
2 The Generic Architecture
for Decision Support Systems
Structurally, a decision support system has four essential components:
a language system (LS)
a presentation system (PS)
a knowledge system (KS)
a problem-processing system (PPS)
These determine its capabilities and behaviors (Bonczek et
al. 1980, 1981a, Dos
Santos and Holsapple 1989, Holsapple and Whinston 1996). The first three are
systems of representation. A language system consists of all messages the DSS
can accept. A presentation system consists of all messages the DSS can emit.
A knowledge system consists of all knowledge the DSS has stored and retained.
By themselves, these three kinds of systems can do nothing, neither individually
nor in tandem. They are inanimate. They simply represent knowledge, either in the
sense of messages that can be passed or representations that have been accumu-
lated for possible future processing.
Although they are merely systems of representation, the KS, LS, and PS are es-
sential elements of a DSS. Each is used by the fourth element: the problem-
processing system. This system is the active component of a DSS. A problem-
processing system is the DSS’s software engine. As its name suggests, a PPS is
what tries to recognize and solve problems (i.
e., process problems) during the
making of a decision. Figure 2 illustrates how the four subsystems of a DSS are
related to each other and to a DSS user. The user is typically a decision maker or
a participant in a decision maker. However, a DSS developer or administrator or
some data-entry person or device could also be a DSS user. In any case, a user
makes a request to the DSS by selecting a desired element of its LS. This could be
a request to accept knowledge, to clarify previous requests or responses, to solve
some problem faced by the decision maker, to detect problems, and so forth. Once
the PPS has been requested to process a particular LS element, it does so. This
processing may very well require the PPS to select some portion of the KS con-
tents, acquire some additional knowledge from external sources (e.
g., a user), or
generate some new knowledge (perhaps using selected or acquired knowledge in
166 Clyde W. Holsapple
doing so). The processing can change the knowledge held in the KS by assimilat-
ing generated or acquired knowledge. The PPS can emit responses to the user by
choosing what PS elements to present.
Thus, some PPS behaviors are overt (witnessed by the user via PPS emission of
PS elements) and others are covert (strictly internal, yielding assimilations of
knowledge into the KS). A problem-processing system does not always have to be
reactive, in the sense of producing behaviors that are reactions to a user’s request.
PPS activity can be triggered by events that are detected inside the DSS or outside
the DSS (Holsapple 1987). For instance, a particular change to the KS content
may trigger some covert or overt PPS behavior such as alerting the user about the
need for a decision about some disturbance or an entrepreneurial opportunity.
Similarly, the acquisition of a particular fact about the DSS’s environment (via
a monitoring device, for example) may trigger overt or covert PPS behavior such
as analysis of a situation’s plausible outcomes with the results being assimilated
into the KS for subsequent use.
The first-order PPS abilities as described above are consistent with previous
characterizations of the generic architecture, but they are also expanded based on
primary knowledge-manipulation activities identified in the collaboratively engi-
neered knowledge-management ontology (Holsapple and Joshi 2002, 2003, 2004).
The five knowledge-manipulation abilities depicted in Figure 2 are the primary,
front-line abilities that comprise a DSS’s contributions to the outcome of a par-
ticular decision episode. These abilities are exercised by the PPS as it works to
find and/or solve problems within a decision process.
The second-order abilities of a PPS shown in Figure 2 are concerned with over-
sight and governance of first-order abilities within and/or across decision episodes.
Figure 2. Basic architecture for decision support systems
DSS Architecture and Types 167
These, too, expand on previous characterizations of the generic DSS architecture
based on the knowledge-management ontology, which identifies coordination, con-
trol, and measurement as important influences on the arrangement and interplay
of the five knowledge manipulation within and across knowledge-management
episodes (Holsapple and Joshi 2000, 2003, 2004). These influences may be whol-
ly exerted by users; or, as Figure 2 indicates, a decision support system’s PPS
may be equipped to govern its own exercise of first-order knowledge-manipulation
abilities.
Coordination refers to a PPS ability of arranging knowledge-manipulation
tasks, and the knowledge flows that connect these tasks, into particular configura-
tions and sequences in the interest of PAIRS results for decision processes. These
manipulation tasks and knowledge flows can be performed by the PPS itself, by
users of the DSS, or a mixture of both computer and human processors. In addi-
tion to governing processing patterns, the coordination ability also involves the
allocation or assignment of particular processors (computer or human) to particu-
lar knowledge-manipulation tasks. Control refers to the ability to ensure the qual-
ity (validity and utility), security, privacy, and sufficiency of knowledge process-
ing that occurs in the course of a decision process in the interest of PAIRS results.
Measurement refers to the ability to track processing and outcomes within and
across decision-making episodes in terms of desired criteria. Such measurements
become a basis for evaluating DSS performance, and perhaps for implementing
adaptive DSSs which are able to improve their behaviors over time based on their
decision support experiences.
As Figure 2 illustrates, the generic DSS architecture recognizes that multiple
types of knowledge may be accommodated within a DSS’s knowledge system.
The most basic of these are descriptive knowledge (often called information),
procedural knowledge, and reasoning knowledge (Holsapple 1995). The first is
knowledge that describes the state of some world of interest. It could be a past
state, present state, future state, expected state, speculative state, and so forth. The
world could be actual, potential, hypothetical, symbolic, fixed, dynamic, physical,
intellectual, emotive, and so forth. In contrast, procedural knowledge characterizes
how to do something (perhaps in one of the worlds of interest). It is a step-wise
specification of what to do in order to accomplish some task or explore some di-
rection. Neither descriptive nor procedural in nature, reasoning knowledge speci-
fies what conclusion is valid when a specific situation is known to exist. It speci-
fies logic that links a premise with a conclusion. The semantics of this linkage can
be varied, including causal, correlative, associative, definitional, advisory, or ana-
logical relationships.
Knowledge of one or more of the three types will exist in the KS of every DSS.
All of this knowledge is susceptible to use by the PPS abilities. The three vertical
bars depicted in the knowledge system of Figure 2 indicate three knowledge orien-
tations that cut across the knowledge types: domain, relational, and self. Know-
ledge oriented toward a domain is the descriptive, procedural, and/or reasoning
knowledge that the PPS uses in grappling with the subject matter of that decision
domain. Relational knowledge is what the decision support system knows about
168 Clyde W. Holsapple
those with whom it interacts. This includes such KS contents as profiles of user
preferences, capabilities, and behaviors, plus knowledge about interpreting LS
elements and picking PS elements. Self knowledge is what the DSS knows about
its own capabilities and behaviors, including KS contents about the structure of
the KS itself and characterizations of what is allowed into the KS via the PPS
assimilation activity. It is fair to say that most DSSs tend to focus on the treatment
of domain knowledge, although the other two knowledge orientations can be very
important from a PAIRS viewpoint.
Figure 2 illustrates a way of organizing the LS and PS contents into subsets
based on the semantics of messages. Some of these subsets may be quite small or
even empty for a particular DSS. Yet another way of categorizing LS requests and
PS responses could be based on distinctions in the styles of messages rather than
differences in their contents. Stylistic distinctions can be quite pronounced, and
a particular DSS may have requests or responses in more than one stylistic cate-
gory. A DSS’s user interface is defined by its LS, its PS, its PPS abilities of know-
ledge acquisition and emission, and its KS contents that the PPS uses for interpret-
ing LS elements and for packaging knowledge into PS elements.
In the generic DSS architecture, we see the crucial and fundamental aspects
common to all decision support systems. To fully appreciate the nature of any
specific decision support system, we must know about the particular requests that
make up its LS, the particular responses that make up its PS, the particular know-
ledge representations allowed (or existing) in its KS, and the particular know-
ledge-processing capabilities of its PPS. If we are ignorant of any of these, then
we cannot claim to have a working knowledge of the DSS. Nor are we in a posi-
tion to thoroughly compare and contrast the DSS with other decision support sys-
tems. Developers of DSSs are well advised to pay careful attention to all four
components when they design and build decision support systems.
3 A Brief History Generic Architecture’s Evolution
The generic architecture for decision support systems began to take shape in the
mid-1970s (Holsapple 1977). In this formative stage, the workings of the problem
processor were emphasized, encompassing such abilities as perception (including
decoding of user requests and finding paths to needed knowledge in the KS), prob-
lem recognition, model formulation, and analysis. It also emphasized the integra-
tion of units of data and modules of procedural knowledge in a computer-based
representation that the problem processor could access. This representation in-
volved extensions to traditional database-management notions, allowing the treat-
ment of both descriptive and procedural knowledge.
Although the ideas of a language system and presentation system were implied
in the early rendition of the architecture, they did not become explicit until later
(Bonczek et
al. 1980, 1981a, Holsapple 1984, Dos Santos and Holsapple 1989).
DSS Architecture and Types 169
The initial work on the framework recognized that a KS could hold (and a PPS
could process) types of knowledge other than the descriptive and procedural varie-
ties. Since then, the possibilities for including reasoning knowledge in a KS have
been explored in greater depth (Bonczek et
al. 1981b, Holsapple 1983, Holsapple
and Whinston 1986, 1996).
The original discussion of the architectural framework advocated incorporation
of artificial intelligence mechanisms into DSSs to produce intelligent decision
support systems. This was further developed (Bonczek et
al. 1979, 1980, 1981a,
1981b, Holsapple and Whinston 1985, 1986) and, today, it is not unusual to find
such mechanisms in the PPSs and KSs of decision support systems.
The original discussion of the architecture emphasized the importance of
knowledge representation and processing in the functioning of a DSS and ad-
vanced the idea of a generalized problem-processing system. This is a PPS that is
invariant across a large array of DSSs and decision-making applications, with all
variations being accommodated by different KSs that all work with the same PPS.
An implementation of this concept appeared in 1983 in the guise of the Know-
ledgeMan (i.
e., the Knowledge Manager) tool for building decision support sys-
tems (Holsapple and Whinston 1983, 1988). This commercial implementation
integrated traditionally distinct knowledge-management techniques into a single
processor that could draw on diverse kinds of objects in a KS (including cells,
fields, variables, text, solvers, forms, charts, menus, and so on) within the context
of a single operation during a problem solving task. Software integration has
become increasingly common. KnowledgeMan was expanded to add a rule man-
agement capability, yielding a generalized PPS (called Guru) for building artifi-
cially intelligent DSSs (Holsapple and Whinston 1986, Osborn and Zickefoose
1990).
With the rise of multiparticipant DSSs, devised to support multiple persons who
engage in making a joint decision, the generic architecture expanded to include
a coordination ability within the PPS, and distinctions between private versus pub-
lic portions of the KS, LS, and PS were identified (Holsapple and Whinston 1996).
Each private segment of the KS is comprised of knowledge representations that are
accessible to only to a particular user (e.
g., a particular participant involved in the
joint decision), whereas the public portion of the KS holds knowledge representa-
tions accessible to all participants in joint decision making. Each private subset of
the LS is comprised of those messages that the PPS will accept only from a particu-
lar participant, whereas all participants can invoke any of the messages in the LS’s
public segment. Similarly, each private subset of the PS is comprised of those mes-
sages that the PPS will emit only to a particular participant, whereas all participants
can view any of the messages in the PS’s public segment.
With the emergence of knowledge management as a field of substantial re-
search over the past decade, the architecture’s PPS second-order abilities have
been further expanded as shown in Figure 2. The architecture allows for meas-
urement and control abilities as described previously. Also, the naming of PPS
first-order abilities has been somewhat adjusted to conform to knowledge-mana-
gement theory.
170 Clyde W. Holsapple
4 DSS Variations
Although the generic architecture gives us a common base and several fundamen-
tal terms for discussing decision support systems, it also lets us distinguish among
different DSSs. For instance, two DSSs could have identical knowledge and pres-
entation systems but differ drastically in their respective language systems. Thus,
the language a user learns for making requests to one DSS may well be of little
use in making requests to the other DSS. The two LSs could vary in terms of the
style and/or content of requests they encompass. Or, they could vary in terms of
dividing lines between public versus private segments.
As another example, two DSSs could have identical PPSs and similar LSs and
PSs. Even the kinds of knowledge representations permitted in their KSs could be
the same. Yet, the two DSSs might exhibit very different behaviors because of the
different knowledge representations actually existing in their KSs. Moreover,
either could exhibit behaviors today it was incapable of yesterday. This situation is
commonly caused by alterations to its KS contents through either the acquisition
or generation of knowledge that is subsequently assimilated. That is, a DSS can
become more knowledgeable.
Even though a relatively generalized problem processor can exist, DSSs can also
differ by having diverse PPSs. All PPSs possess the first-order abilities of acquisi-
tion, selection, assimilation, and emission. Many have a knowledge-generation
ability too. The exact character of each ability can differ widely from one problem-
processing system to the next. For example, the selection and generation abilities of
one PPS may be based on a spreadsheet technique for managing knowledge,
whereas that technique is entirely absent from some other PPS that emphasizes
database-management or rule-management techniques for handling knowledge.
This implies KS differences as well. When a PPS employs a spreadsheet processing
approach, the DSS’s knowledge system uses a corresponding spreadsheet approach
to knowledge representation. In contrast, if a DSS’s problem processor relies on
a database-management technique for processing knowledge, then its KS must
contain knowledge represented in terms of databases. In other words, DSSs can
differ with respect to the knowledge-management techniques with which their PPSs
are equipped and that govern the usable representations held in their KSs.
Many special cases of the generic DSS architecture can be identified by view-
ing KS contents and PPS abilities is in terms of the knowledge-management tech-
niques employed by a DSS. Each technique characterizes a particular class of
decision support systems by:
restricting KS contents to representations allowed by a certain know-
ledge-management techniques(s), and
restricting the PPS abilities to processing allowed by the technique(s).
The result a specialized architecture with the generic traits suggested in Figure 2,
but specializing in a particular technique or techniques for representing and pro-
cessing knowledge (Holsapple and Whinston 1996).
DSS Architecture and Types 171
For example, a special class of decision support systems uses the spreadsheet
technique of knowledge management. The KS of each DSS in this class consists
of descriptive and procedural knowledge represented in a spreadsheet fashion. The
PPS of such a DSS consists of software that can acquire knowledge for manipulat-
ing these representations, select or generate knowledge from them, and present
them in a form understandable to users. In contrast, the DSS that uses a database-
management technique has very different representations in its KS, and it has
a PPS equipped to process them rather than providing spreadsheet representations.
Although both spreadsheet and database DSS classes adhere to the generic archi-
tecture, each can be viewed in terms of its own more specialized framework.
Several of the more common specialized frameworks are examined here: text,
hypertext, database, spreadsheet, solver, expert system, and compound frameworks.
Each characterizes a type or class of decision support systems. Many other DSS
types are conceivable. Most tend to emphasize one or two knowledge-management
techniques for representing KS contents and defining PPS behaviors. As we intro-
duce these special cases of the generic architecture, we also present broad outlines
of corresponding knowledge-management techniques that they employ. Subsequent
chapters examine many of these types of DSSs in greater detail.
4.1 Text-Oriented Decision Support Systems
For centuries, decision makers have used the contents of books, periodicals, let-
ters, and memos as textual repositories of knowledge. In the course of decision
making, their contents are available as raw materials for the manufacturing pro-
cess. The knowledge embodied in a piece of text might be descriptive, such as
a record of the effects of similar decision alternatives chosen in the past, or a de-
scription of an organization’s business activities. It could be procedural know-
ledge, such as passage explaining how to calculate a forecast or how to acquire
some needed knowledge. The text could embody reasoning knowledge, such as
rules of thumb indicating likely causes of or remedies for an unwanted situation.
Whatever its type, the decision maker searches and selects pieces of text to be-
come more knowledgeable, to verify impressions, or to stimulate ideas.
In the 1970s and especially in the 1980s, text management emerged as an im-
portant, widely used computerized means for representing and processing pieces
of text. Although its main use has been for such clerical activities (preparing and
editing letters, reports, and manuscripts, for instance), it can also be of value to
decision makers (Keen 1987, Fedorowicz 1989). The KS of a text-oriented DSS is
made up of electronic documents, each being a textual passage that is potentially
interesting to the decision maker.
The PPS consists of software that can perform various manipulations on contents
of any of the stored documents. It may also involve software that can help a user in
making requests. The LS contains requests corresponding to the various allowed
manipulations. It may also contain requests that let a user ask for assistance covering
172 Clyde W. Holsapple
some aspect of the DSS. The PS consists of images of stored text that can be emitted,
plus messages that can help the decision maker use the DSS.
An example will help illustrate the value of text-oriented DSSs to decision
makers. Imagine that you are a product manager concerned with ensuring the
success of a technically complex product. A number of the many decisions you
face involve deciding about what features the product should have. Such decisions
depend on many pieces of knowledge. Some tell about the technical feasibility of
features, whereas others indicate research, development, and production costs
associated with various features. You need to know about the features offered by
competing products. How would potential customers assess the cost-benefit trade-
offs of specific features? What legal, health, safety, and maintenance issues must
you consider for each potential feature?
During the course of each week, you get an assortment of product ideas that de-
serve to be checked out when you get the time – if only you could remember all of
them. With a text-oriented DSS, you keep electronic notes about the ideas as they
arise, which consists of typing in the text that you want the PPS to assimilate into
its KS. You might put all the ideas in a single, large file of text. Or, it may be
more convenient to organize them into multiple text files and folders (e.
g., ideas
about different features are stored as text in different files). You may want to ex-
pand on an idea that has been stored in the KS. To do so, you use the LS to select
the document holding that idea and to revise it. If an idea needs to be discarded,
then the corresponding text is deleted instead of revised.
Suppose you want to make a decision about the inclusion or nature of some
product feature. The stored text containing ideas about that feature can be selected
from the KS, emitted for viewing on a console or on paper. Rather than rummag-
ing through the selected text, you may want to restrict your attention to only those
ideas concerned with the cost of the feature. Then, you use the LS to indicate that
the PPS should select text having the “cost” keyword and emit its surrounding text
for display. The KS can hold other pieces of text entered by an assistant who col-
lects and summarizes information about features of competing products, for as-
similation into the KS. Selection via focused searching or browsing through such
text may also support your efforts at reaching the feature decision.
Traditional text-management software does little in the way of generating know-
ledge that could be applied to support a decision. However, generating knowledge
from text is becoming increasingly important through such functionalities as text
mining (Nasukawa and Nagano 2001, Froelich et
al. 2005) and content analysis
(Neundorf 2004).
4.2 Hypertext-Oriented Decision Support Systems
In general, a text-oriented DSS supports a decision maker by electronically keep-
ing track of textually represented knowledge that could have a bearing on deci-
sions. It allows documents to be electronically created, revised, and reviewed by
DSS Architecture and Types 173
a decision maker on an as-needed basis. The viewing can be exploratory browsing
in search of stimulation or a focused search for some particular piece of know-
ledge needed in the manufacture of a decision. In either event, there is a problem
with traditional text management: it is not convenient to trace a flow of ideas
through separate pieces of text. There is no explicit relationship or connection
between the knowledge held in one text file and the knowledge in another.
This problem is remedied by a technique known as hypertext. Each piece of
text is linked to other pieces of text that are conceptually related to it. For instance,
there may be a piece of text about a particular competitor. This text can be linked
to pieces of text about other competitors. It can also be connected to each piece of
text that discusses a feature offered by that competitor. It is probably associated
with still other text summarizing current market positions of all competing pro-
ducts. This summary, is in turn, linked to text covering the results of market sur-
veys, to a narrative about overall market potential, to notes containing marketing
ideas, and so forth.
In addition to the PPS capabilities of a traditional text-oriented DSS, a user can
request the creation, deletion, and traversal of links. In traversing a link, the PPS
shifts its focus (and the user’s) from one piece of text to another. For instance,
when looking at market-summary text, you want more information about one of
the competitors noted there. Thus, you request that the PPS follow the link to that
competitor’s text and display it to you. In examining it, you see that it is linked to
one of the features that is of special interest to you. Requesting the PPS to follow
that link brings a full discussion of that feature into view. Noting that it is con-
nected to another competitor, you move to the text for that competitor. This ad hoc
traversal through associated pieces of text continues at your discretion, resembling
a flow of thoughts through the many associated concepts in your own mind.
The benefit of this hypertext kind of DSS is that it supplements a decision
maker’s own capabilities by accurately storing and recalling large volumes of
concepts and connections that he or she is not inclined personally to memorize
(Minch 1989, Bieber 1992, 1995).
With the advent and explosive growth of the World Wide Web, hypertext rep-
resentations of knowledge are so commonplace that they are often taken for
granted. Web-oriented DSSs comprise a large portion of the class of hypertext-
oriented DSSs. Indeed, the World Wide Web can be viewed as a vast distributed
KS, whose PPS is also distributed – having a local browser component and remote
components in the guise of server software (Holsapple et
al. 2000).
4.3 Database-Oriented Decision Support Systems
Another special case of the DSS architecture consists of those systems developed
with the database technique of knowledge management. Although there are several
important variants of this technique, perhaps the most widely used is relational
database management. It is the variant we consider here. Rather than treating data
174 Clyde W. Holsapple
as streams of text, they are organized in a highly structured, tabular fashion. The
processing of these data tables is designed to take advantage of their high degree of
structure. It is, therefore, more intricate than text processing.
People have used database management in decision support systems using since
the early years of the DSS field (e.
g., Bonczek et
al. 1976, 1978, Joyce and Oliver
1977, Klass 1977). Like text-oriented DSSs, these systems aid decision makers by
accurately tracking and selectively recalling knowledge that satisfies a particular
need or serves to stimulate ideas. However, the knowledge handled by database-
oriented DSSs tend to be primarily descriptive, rigidly structured, and often ex-
tremely voluminous.
The computer files that make up its KS hold information about table structures
(e.
g., what fields are involved in what tables) plus the actual data value contents
of each table. The PPS has three kinds of software: a database control system, an
interactive query processing system, and various custom-built processing systems.
One – but not both – of the latter two could be omitted from the DSS. The data-
base control system consists of capabilities for manipulating table structures and
contents (e.
g., defining or revising table structures, finding or updating records,
creating or deleting records, and building new tables from existing ones). These
capabilities are used by the query processor and custom-built processors in their
efforts at satisfying user requests.
The query processing system is able to respond to certain standard types of re-
quests for data retrieval (and perhaps for help). These requests comprise a query
language and make up part of the DSS’s language system. Data-retrieval requests
are stated in terms of the database’s structure. They tell the query processor the
fields and tables for which the user is interested in seeing data values. The query
processor then issues an appropriate sequence of commands to the database con-
trol system, causing it to select the desired values from the database. These values
are then presented in some standard listing format (an element of the PS) for the
user to view.
For a variety of reasons, users may prefer to deal with custom-built processors
rather than standard query processors. They may give responses more quickly to
requests a standard query could not handle, presenting responses in a specially
tailored fashion without requiring the user to learn the syntax of a query language
or to use as many keystrokes. A custom-built processor might be built by the
DSS’s user but is more likely to be constructed by someone who is well versed in
computer science. Such a processor is often called an application program, be-
cause it is a program that has been developed to meet the specific needs of a mar-
keting, production, financial, or other application.
Embedded within a custom-built processor program is the logic to interpret
some custom-designed set of requests. In such an application program, there will
be commands to the database control system, telling it what database manipula-
tions to perform for each possible request. There will also be the logic necessary
for packaging responses in a customized manner. There may even be some calcu-
lations to generate new knowledge based on values from the database. Calculation
DSS Architecture and Types 175
results can be included in an emitted response and/or assimilated into the KS for
subsequent use.
By the 1990s, a special class of database systems known as data warehouses had
emerged. A data warehouse is a large collection of data integrated from multiple
operational systems, oriented toward a particular subject domain, whose content is
not over-written or discarded, but is time-stamped as it is assimilated (Inmon 2002).
A data warehouse may have the look of a relational database or a multidimensional
database (Datta and Thomas 1999). Data warehouse technology was specifically
conceived to devise KSs for high-performance support of decision-making proc-
esses.
4.4 Spreadsheet-Oriented Decision Support Systems
In the case of a text-oriented DSS, procedural knowledge can be represented in
textual passages in the KS. About all the PPS can do with such a procedure is
display it to the user and modify it at the user’s request. It is up to the user to carry
out the procedure’s instructions, if desired. In the case of a database-oriented DSS,
extensive procedural knowledge cannot be readily represented in the KS. How-
ever, the application programs that form part of the PPS can contain instructions
for analyzing data selected from the database. By carrying out these procedures,
the PPS can emit new knowledge (e.
g., a sales forecast) that has been generated
from KS contents (e.
g., records of past sales trends). But, because they are part of
the PPS, a user cannot readily view, modify, or create such procedures, as can be
done in the text-oriented case.
Using the spreadsheet technique for knowledge management, a DSS user not
only can create, view, and modify procedural knowledge assimilated in the KS,
but also can tell the PPS to carry out the instructions they contain. This gives DSS
users much more power in handling procedural knowledge than is achievable with
either text management or database management. In addition, spreadsheet man-
agement is able to deal with descriptive knowledge. However, it is not nearly as
convenient as database management in handling large volumes of descriptive
knowledge, nor does it allow a user to readily represent and process data in textual
passages.
In a spreadsheet-oriented DSS, the knowledge system is comprised of files that
house spreadsheets, each being a grid of cells. It may be a small grid, involving
only a few cells, or very large, encompassing hundreds (or perhaps thousands) of
cells. Each cell has a unique name based on its location in the grid. In addition to
its name, each cell can have a definition and a value. A cell definition tells the PPS
how to determine that cell’s value. There are two common kinds of cell defini-
tions: constants and formulas. The value of a cell defined as a constant is merely
the constant itself. In contrast, a formula contains names of other cells, perhaps
some constants, and some operators or functions indicating how to combine the
values of named cells and constants. The result of this calculation becomes the
value of the cell having a formula definition.
176 Clyde W. Holsapple
Taken together, the formulas of a spreadsheet constitute a chunk of procedural
knowledge, containing instructions that the PPS can carry out to generate new
knowledge. The results of performing this procedure are cell values of interest to
the user. Spreadsheet-oriented DSSs are typically used for what-if analyses in
order to see the implications of some set o assumptions embodied in the cell defi-
nitions. They support a decision maker by giving a rapid means of revaluating
various alternatives. Today, spreadsheet-oriented DSSs are heavily used in organi-
zations (McGill and Koblas 2005).
In addition to holding procedural knowledge (in the guise of formula cells) and
descriptive knowledge (in the guise of numeric constant cells), a spreadsheet file
can also hold some simple presentation knowledge and linguistic knowledge.
When specifying a spreadsheet, a user can define some cells as string constants
(e.
g., “Sales”) to show up as labels, titles, and explanations when the spreadsheet
is displayed. This presentation knowledge makes the results of calculations easier
to grasp.
Conversely, a user’s task in making a request (especially to define cells) can be
eased by macros. A macro is a name (usually short) the user can define to corre-
spond to an entire series of keystrokes. The macro and its meaning are stored as
linguistic knowledge in a spreadsheet file, effectively extending the LS. For in-
stance, the macro name D might be defined to mean the keystrokes D5*D6D7. In
subsequent requests, macro names such as D can be used instead of the lengthy
series of keystrokes they represent. To interpret such a request, the PPS finds the
meaning of the macro name in its KS.
4.5 Solver-Oriented Decision Support Systems
Another special class of decision support systems is based on the notion of
solvers. A solver is a procedure consisting of instructions that a computer can
execute in order to solve any member of a particular class of problems. For in-
stance, one solver might be able to solve depreciation problems. Another might be
designed to solve portfolio analysis problems. Yet another might solve linear op-
timization problems. Solver management is concerned with the storage and use of
a collection of solvers (Holsapple and Whinston 1996, Lee and Huh 2003).
A solver-oriented DSS is frequently equipped with more than one solver, and
the user’s request indicates which is appropriate for the problem at hand. The
collection of available solvers is often centered around some area of problems
such as financial, economic, forecasting, planning, statistical, or optimization
problems. Thus, one DSS might specialize in solving financial problems; another
has solvers to help in various kinds of statistical analysis; and yet another might do
both of these.
There are two basic approaches for incorporating solvers into a DSS: fixed and
flexible. In the fixed approach, solvers are part of the PPS, which means that a sol-
ver cannot be easily added to or deleted from the DSS nor readily modified. The
set of available solvers is fixed, and each solver in that set is fixed. About all
DSS Architecture and Types 177
a user can choose to do is execute any of the PPS solvers. This ability may be
enough for many users’ needs. However, other users may need to add, delete,
revise and combine solvers over a lifetime of a DSS. With this flexible approach,
the PPS is designed to manipulate (e.
g., create, delete, update, combine, coordi-
nate) solvers according to user requests. First, consider the fixed approach in a bit
more detail, and then do the same for the flexible approach.
In the fixed solver case, the PPS commonly has the ability to acquire, assimilate,
select, and emit descriptive knowledge in the KS in the form of data sets, problem
statements, and/or report templates. A data set is a parcel of descriptive knowledge
that can be used by one or more solvers in the course of solving problems. It usu-
ally consists of groupings or sequences of numbers organized according to conven-
tions required by the solvers. For example, we may have used PPS to assimilate
a data set composed of revenue and profit numbers for each of the past 15 years.
This data set could be used by a basic statistics solver to give the average and stan-
dard deviation of revenues and profits. The same data set could be used by a fore-
casting solver to produce a prediction of next year’s profit, assuming a certain
revenue level for the next year. Using a different data set, this same solver could
produce a forecast of sales for an assumed level of advertising expenditures. Thus,
many solvers can use a data set, and a given solver can feed on multiple data sets.
In addition to data sets, it is not uncommon for this kind of DSS to hold prob-
lem statements and report format descriptions in its KS. Because the problem
statement requests permitted by the LS can be very lengthy, fairly complex, and
used repeatedly, it can be convenient for a user to edit them (i.
e., create, recall,
revise them), much like pieces of text. Each problem statement is an LS element
that indicates the solver and mode of presentation to be used in printing or display-
ing the solution. The latter may designate a standard kind of presentation (e.
g.,
a pie graph with slice percentages shown) or a customized report. The format of
such a report is something the user specifies. Details of this specification can be-
come quite lengthy and, therefore, are convenient to store as presentation know-
ledge in the KS. This knowledge defines a portion of the PS.
The flexible approach to handling solvers in a DSS also accommodates data
sets and perhaps problem statements or report formats in its KS. But, the KS holds
solver modules as well. A solver module is procedural knowledge that the PPS can
execute to solve a problem. Each module requires certain data to be available for
its use before its instructions can be carried out. Some of that data may already
exist in KS data sets. The remaining data must either be furnished by the user
(i.
e., in the problem statement) or generated by executing other modules. In other
words, a single module may not be able to solve some problems. Yet, they can be
solved by executing a certain sequence of modules (Bonczek et
al. 1981b). Results
of carrying out instructions in the first module are used as data inputs in executing
the second module, whose results become data for the third or subsequent module
executions, and so forth, until a solution is achieved. Thus, a solver can be formed
by combining and coordinating the use of available modules so that the data out-
puts of one can be data inputs to another.
178 Clyde W. Holsapple
The LS contains problem statements as well as requests that let a user edit KS
contents. It may also contain requests for assistance in using the system. In a prob-
lem statement, the user typically indicates which module or module sequence is to
be used in addressing the problem. It may also specify some data to serve as mod-
ule inputs or identify data sets as module inputs. Upon interpreting such a request,
the PPS selects the appropriate module or modules from the KS. With some DSSs
of this kind, the PPS is able to select modules that are implied (but not explicitly
identified) in the problem statement or to combine modules into a proper sequence
without being told a definite sequence in the problem statement. This capability
may be rely on KS reasoning knowledge about what solver module to use in each
given situation.
By bringing a copy of a selected module into its working memory, the PPS is
able to carry out the procedure of instructions it contains. The input data it needs
to work on and the output data it generates are also kept in working memory while
the module is being executed. After the PPS is finished executing a module, its
instructions and any data not needed by the next module to be executed are elimi-
nated from the PPS’s working memory. They are replaced by a copy of the next
module and data inputs it needs. The PPS may need to restructure data produced
by formerly executed modules so it can be used by the module that is about to be
executed. Thus, the PPS coordinates the executions of modules that combine to
make up the solver for a user’s problem statement.
The LS requests that a user employs to edit KS contents mirror corresponding
PPS capabilities. In broad terms, they allow users to create, revise, and delete
modules or data sets (and perhaps report templates or problem statements as well).
In creating a new module, for instance, a user would specify the instructions that
make up this piece of procedural knowledge. Typically, this is done in much the
same way that text is entered when using a text-management technique. However,
the instructions are stated in a special language (e.
g., programming language) that
the PPS can understand and, therefore, carry out during the module execution.
Assimilating a new module into the KS can also involve facilities for testing it to
ensure that it produces correct results and for converting the module to an equiva-
lent set of instructions that the PPS can process more efficiently.
As in the fixed approach, a flexible solver-oriented DSS may allow users to re-
quest a customized presentation of solver results. The desired formatting can be
specified as part of the problem statement request. Alternatively, it could be stored
in the KS as a template that can be revised readily and used repeatedly by simply
indicating its name in problem statements.
4.6 Rule-Oriented Decision Support Systems
Another special case of the generic DSS architecture is based on a knowledge-
management technique that involves representing and processing rules. This tech-
nique evolved within the field of artificial intelligence, giving computers the ability
to manage reasoning knowledge. Recall that reasoning knowledge tells us what
DSS Architecture and Types 179
conclusions are valid when a certain situation exists. Rules offer a straightforward,
convenient means for representing such fragments of knowledge. A rule has the
basic form
If: description of a possible situation (premise)
Then: indication of actions to take (conclusion)
Because: justification for taking those actions (reason)
This format says that if the possible situation can be determined to exist, then the
indicated actions should be carried out for the reasons given. In other words, if the
premise is true, then the conclusion is valid.
The KS of a rule-oriented DSS holds one or more rule sets, where each rule set
pertains to reasoning about what recommendation to give a user seeking advice on
some subject (Holsapple and Whinston 1986). For instance, one set of rules might
be concerned with producing advice about correcting a manufacturing process that
is turning out defective goods. Another rule set might hold reasoning knowledge
needed to produce recommendations about where to site additional retail outlets.
Yet another rule set could deal with portfolio advice sought by investors. In addi-
tion to rule sets, it is common for the KS to contain descriptions of the current state
of affairs (e.
g., current machine settings, locations of competing outlets, an inves-
tor’s present financial situation). Such state descriptions can be thought of as val-
ues that have been assigned to variables.
Aside from requests for help and for editing state descriptions, users of a rule-
oriented DSS can issue two main types of requests for decision support purposes.
The LS contains requests for advice and requests for explanation. For example, in
making a decision about what corrective action to take, the decision maker may
request the DSS to advise him or her about the likely causes of cracks in a metal
part. The decision maker may subsequently request an explanation of the rationale
for that advice. Correspondingly, the PS includes messages presenting advice and
explanations.
The problem processor for a rule-oriented DSS has capabilities for creating, re-
vising, and deleting state descriptions. Of greater interest is the capability to do
logical inference (i.
e., to reason) with a set of rules to produce advice sought by
a user. The problem processor examines pertinent rules in a rule set, looking for
those whose premises are true for the present situation. This situation is defined by
current state descriptions (e.
g., machine settings) and the user’s request for advice
(e.
g., citing the nature of the quality defect). When the PPS finds a true premise, it
takes the actions specified in that rule’s conclusion. This action sheds further light
on the situation, which allows premises of still other rules to be established as true,
causing actions in their conclusions to be taken. Reasoning continues in this way
until some action is taken that yields the requested advice or the PPS gives up due
to insufficient knowledge in its KS. The PPS also has the ability to explain its
behavior both during and after conducting the inference. There are many possible
variations for the inference process for both the forward reasoning approach just
outlined and the alternative reverse-reasoning approach which involves goal-
seeking (Holsapple and Whinston 1986, 1996).
180 Clyde W. Holsapple
A rule-oriented DSS is also known as an expert system because it emulates the
nature of a human expert from whom we may seek advice in the course of making
a decision (Bonczek et
al. 1980, Holsapple and Whinston 1986). This special kind
of DSS is particularly valuable when human experts are unavailable, too expen-
sive, or perhaps erratic. Rather than asking the human expert for a recommenda-
tion and explanation, the expert system is asked. Its rule sets are built to embody
reasoning knowledge similar to what its human counterpart uses. Because its in-
ference mechanisms process those rules using basic principles of logic, the PPS
for this kind of decision support system is often called an inference engine.
An expert system is always available for consultation: 24 hours per day, seven
days per week, year-round. It does not charge high fees every time it is consulted.
It is immune to bad days, personality conflicts, political considerations, and over-
sights in conducting inference. To the extent that its reasoning and descriptive
knowledge is not erroneous, it can be an important knowledge source for decision
makers.
4.7 Compound Decision Support Systems
Each of the foregoing special cases of the generic DSS framework has tended to
emphasize one knowledge-management technique, be it text, hypertext, database,
spreadsheet, solver, or rule management. Each supports a decision maker in ways
that cannot be easily replicated by a DSS oriented toward a different technique. If
a decision maker would like the kinds of support offered by multiple knowledge-
management techniques, there are two basic options:
Use multiple DSSs, each oriented toward a particular technique
Use a single DSS that encompasses multiple techniques
Some decision makers prefer the first option. Others prefer the second.
The first option is akin to having multiple staff assistants, each of whom is well
versed in a single knowledge-management technique. One is good at representing
and processing text, another at handling solvers, another at managing rules, and so
forth. Each has its own LS and PS, which the decision maker must learn in order
to make requests and appreciate responses. When results of using one technique
need to be processed via another technique, it is the decision maker’s responsibil-
ity to translate responses from one DSS into requests to another DSS. For in-
stance, a solver-oriented DSS might produce an economic forecast that a rule-
oriented DSS needs to consider when reasoning about where to locate a new retail
outlet.
There are several approaches to integration across DSSs: conversion, clipboard,
and confederation (Holsapple and Whinston 1984, 1996). Conversion requires
a facility that can convert outputs of one PPS into a form that is acceptable as
input to another PPS. This can be a piece of software separate from the PPSs.
Alternatively, it can be built into the acquisition or emission ability of a PPS, as an
import/export functionality that can accept knowledge representations emitted by
DSS Architecture and Types 181
an alien PPS or package emissions into representations that an alien PPS can in-
terpret. With the clipboard approach, transferal of knowledge between processors
involves an intermediary repository (i.
e., clipboard) having a format that each PPS
can both copy knowledge into and grab knowledge from. For a confederation, the
task of copying to and pasting from a clipboard is eliminated. Instead, all of the
confederated PPSs share a common KS, having a single knowledge representation
format that can be directly interpreted by each of the distinct PPSs. Each of these
three approaches accomplishes integration by working on achieving commonal-
ity/compatibility of knowledge representation, while maintaining distinct proces-
sors and processing capabilities.
The second option is akin to having a staff assistant who is adept at multiple
knowledge-management techniques. There is one LS and one PS for the decision
maker to learn. Although they are probably more extensive than those of a particu-
lar single-technique DSS, they are likely less demanding than coping with the sum
total of LSs and PSs for all corresponding single-technique DSSs. The effort re-
quired of a decision maker who wants to use results of one technique in the proc-
essing of another technique varies, depending on the way in which the multiple
techniques have been integrated into a single compound DSS.
There are two main approaches to integration within a DSS: nesting and syn-
ergy (Holsapple and Whinston 1984, 1996). In the nested approach, a traditionally
separate knowledge-management technique is nested within the capabilities of
another. For instance, solver management can be found nested within spreadsheet
management (e.
g., Microsoft Excel) and spreadsheet management can be found
nested within text management (e.
g., Microsoft Word). A nested technique typi-
cally does not have features that are as extensive as those found in a standalone
tool dedicated to that technique. Moreover, the nested capabilities cannot be read-
ily used by persons unfamiliar with the dominant technique. However, there is no
need to switch back and forth among distinct processors (i.
e., there is a single
PPS) and there are not multiple knowledge systems whose contents need to be
consistently maintained. Thus, from a DSS standpoint, nesting results in a tool that
can function as a single PPS having multiple knowledge-processing capabilities
that are accessible via a unified LS and PS when operating on a single KS.
In the synergistic approach to integrating traditionally distinct knowledge-
management techniques, there is no nesting, no dominant technique, and no sec-
ondary nested techniques. All techniques are integrated into a single tool that
allows any capability to be used independently of another, or together with an-
other within a single operation. For instance, a database operation can be per-
formed without knowing about spreadsheets, and vice versa; but the same tool can
satisfy a request to select database contents conditional on values of spreadsheet
cells or define a cell’s value in terms of run-time retrieval from a database
(Holsapple and Whinston 1988). When a synergistically integrated tool is adopted
as the PPS of a decision support system, the KS is comprised of many kinds of
objects: cells, spreadsheets, macros, fields, database tables, variables, solver mo-
dules, presentation templates, text, rules, charts, programs, and so forth. The PPS
182 Clyde W. Holsapple
can utilize any of these knowledge representations as it works to satisfy a user’s
request or respond to an event.
Figure 3 portrays an example of a DSS whose PPS is synergistically integrated.
With this kind of integration, the dividing lines between traditional knowledge-
management techniques become blurred. There is a KS with multiple kinds of
objects, some encapsulating representations of descriptive knowledge, others pro-
cedural knowledge, and others reasoning knowledge. The PPS can manipulate any
of these representations through its first-order abilities: acquiring, assimilating,
selecting, generating, emitting knowledge represented as text objects, database
objects, spreadsheet objects, solver objects, and so on.
The LS consists primarily of requests for assistance and knowledge manipula-
tion. The former allow a user to ask for help in issuing requests or clarification of
DSS responses. A knowledge-manipulation request could look very much like
standard requests made to single-technique DSSs – that is, it deals with only one
technique (e.
g., spreadsheet) and the user is expected to understand that technique
(e.
g., the notion of cell definitions). Other knowledge-manipulation requests may
not require such understanding and may even trigger sequences of PPS manipu-
lations involving multiple techniques. For instance, the LS may allow a user to
type: Show revenue projection for region = “south” or to pick a comparable op-
tion from a menu. This request is not necessarily oriented toward any particular
technique. The PPS interprets it as meaning that certain data need to be selected
from a KS database, that a rule set is to be used to generate an appropriate se-
quence of solvers via inference, and that those selected solvers are then to be exe-
cuted with selected data in order to generate the revenue projection. The user does
Figure 3. Example of a compound DSS with synergistic integration
DSS Architecture and Types 183
not need to know about database, rule set, or solver manipulations. These activi-
ties happen beneath the customized DSS surface provided by the PPS.
Manipulation or assistance requests and responses may be standardized or cus-
tomized for a specific user. Customization can be based on relational knowledge
held in the KS. This relational knowledge can profile a user in descriptive, proce-
dural, and/or reasoning terms to permit customized packaging of emitted re-
sponses or customized interpretation of acquired messages. For instance, the pre-
viously requested revenue projection, along with explanatory commentary, might
be presented in a multicolor form that is personalized (e.
g., showing the user’s
name and request date) and in which the projected number blinks if it exceeds last
year’s revenue by more than 20%. Specifications of colors and arrangements of
items in the form would exist as presentation knowledge in the KS.
We close the overview of compound DSSs with a combination of the flexible
solver and database-management techniques from Sprague and Carlson (1982) as
shown in Figure 4. In this special case of compound DSSs, the KS is comprised of
a database and a model base. The term model base refers to solver modules exist-
ing in the KS (Lee and Huh 2003). Correspondingly, the PPS includes database-
management software for manipulating the database portion of the KS and model-
base-management software for manipulating the KS’s model base. Executing
a solver with selected database contents generates new knowledge for the user.
The dialog generation and management system is that part of the PPS that inter-
prets user requests, providing help, and presenting responses. Although LS and PS
components of a DSS are not explicit in Figure 4, they are implicit in the notion of
a dialog and have, respectively, been referred to as an action language and display
language (Sprague and Carlson 1982).
Figure 4. Combining database and solver techniques in a compound DSS
184 Clyde W. Holsapple
The framework shown in Figure 4 is often cited in DSS books and articles as
the architecture for decision support systems (Sprague and Carlson 1982, Thierauf
1988, Turban 1988). However, it covers only a portion of DSS possibilities identi-
fied by the generic architecture. Nevertheless, it is an important special case of
that architecture that stresses the combination of database and solver techniques
for knowledge management. A variation of this combination is heavily used in
large organizations today: combining a data warehouse with analytical solvers
(called online analytical processing) to derive new knowledge (Chaudhuri and
Dayal 1997, Koutsoukis et
al. 1999). A further variation combines a data ware-
house with data-mining solvers that generate knowledge by discovering patterns in
data that are helpful in decision making (Han and Kamber 2000).
4.8 Multiparticipant Decision Support Systems
A decision maker can have multiple participants who contribute to the making of
a decision. Some or all of these participants may share authority over the decision.
There can be some participants who have no authority over the decision, but do
wield influence over what the decision will be. When a computer-based system
supports a multiparticipant decision maker (be it a group, team, or organization),
we call it a multiparticipant decision support system (MDSS). The DSS architec-
ture shown in Figure 2 encompasses this situation.
MDSSs that support group decision making have developed and matured over
a period many years (Gray and Nunamaker 1993). They have been the subject of
much research (Fjermestad and Hiltz 2000, Fjermestad 2004) and there are many
examples of their successful application (e.
g., Nunamaker et
al. 1989, Adkins
et
al. 2003). The hallmark of many group decision support system implementa-
tions is a PPS that has a strong coordination ability for handling or even guiding
participant interactions (Turoff 1991), coupled with first-order abilities of acquir-
ing knowledge from participants, assimilating this knowledge into the KS which
functions as a group memory, and selecting and emitting KS contents to partici-
pants. Both private and public sectors of the KS exist.
Another kind of MDSS concentrates on supporting a negotiation among par-
ticipants. The outcome of the negotiation is a decision on which participants
agree. Increasingly, these kinds of MDSSs are supporting negotiations over the
World Wide Web (Kersten and Noronha 1999). Negotiation support systems tend
to have PPSs with fairly well developed second-order abilities of coordination
and control. Perhaps the most extensive second-order abilities belong to the
problem-processing systems of MDSSs intended to support participants organ-
ized into relatively complex structures of authority, influence, specialization, and
communication: enterprises, supply chains, large project teams, markets. Re-
search examining these organizational decision support systems is quite varied
(e.
g., George et
al. 1992, Rein et
al. 1993, Santhanam et
al. 2000, Kwan and
Balasubramanian 2003, Holsapple and Sena 2005), but not yet as voluminous as
DSS Architecture and Types 185
that focused on group decision support systems. The architecture shown in
Figure 2 furnishes a framework for guiding future progress in understanding the
nature, possibilities, and outcomes of these kinds of multiparticipant decision
support systems. The current state of the art for MDSSs is covered in a series of
chapters later in this volume.
5 Summary
This chapter introduces the generic DSS architecture. 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 represen-
tation: 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 prob-
lem processor is a dynamic system that can accept any request in the LS and react
with a corresponding response from the PS. The response corresponding to
a 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. Some DSSs can even produce responses without having
received a corresponding request. In addition to reacting to users, they take initia-
tive in the processing of knowledge, reacting to events.
There are many special cases of the generic DSS architecture, each charac-
terizing a distinct class of decision support systems. Several of these specialized
frameworks have been examined here. They differ due to their emphasis on one or
another popular knowledge-management technique. This examination of special-
ized cases serves several purposes. First, it reinforces an understanding of the
generic architecture by illustrating what it is meant by a KS, PPS, LS, and PS.
Second, it offers an overview of important kinds of DSSs. Third, it gives a brief
introduction to several of the key classes of DSSs that receive more in-depth co-
verage in ensuing chapters. Fourth, it provides a useful background for thinking
about issues that face the developers of decision support systems.
Acknowledgements
Some portions of this chapter have been reproduced, with permission, from Decision
Support Systems: A Knowledge-Based Approach, C. W. Holsapple and A. B. Whin-
ston, St. Paul: West, 1996.
186 Clyde W. Holsapple
References
Adkins, M., M. Burgoon and J.F. Nunamaker, “Using Group Support Systems for
Strategic Planning with the United States Air Force,” Decis Support Syst, 34(3),
2003.
Bieber, M., “Automating Hypermedia for Decision Support,” Hypermedia, 4(2),
1992.
Bieber, M., “On Integrating Hypermedia into Decision Support and Other Infor-
mation Systems,” Decis Support Syst, 14(3), 1995.
Bonczek, R.H., C.W. Holsapple and A.B. Whinston, “A Decision Support System
for Area-Wide Water Quality Planning,” Socio Econ Plan Sci, 10(6), 1976.
Bonczek, R.H., C.W. Holsapple and A.B. Whinston, “Aiding Decision Makers
with a Generalized Database Management System,” Decision Sci, April, 1978.
Bonczek, R.H., C.W. Holsapple and A.B. Whinston, “The Integration of Data
Base Management and Problem Resolution,” Inform Syst, 4(2), 1979.
Bonczek, R.H., C.W. Holsapple and A.B. Whinston, “Future Directions for De-
veloping Decision Support Systems,” Decision Sci, October, 1980.
Bonczek, R.H., C.W. Holsapple and A.B. Whinston, Foundations of Decision
Support Systems. New York: Academic, 1981a.
Bonczek, R.H., C.W. Holsapple and A.B. Whinston, “A Generalized Decision
Support System Using Predicate Calculus and Network Data Base Manage-
ment,” Oper Res, 29(2), 1981b.
Chaudhuri, S. and U. Dayal, “An Overview of Data Warehousing and OLAP
Technology,” ACM SIGMOD Rec, 26(1), 1997.
Datta, A. and H. Thomas, “The Cube Data Model: A Conceptual Model and Al-
gebra for On-line Analytical Processing in Data Warehouses,” Decis Support
Syst, 27(3), 1999.
Dos Santos, B. and C.W. Holsapple, “A Framework for Designing Adaptive DSS
Interfaces,” Decis Support Syst, 5,(1), 1989.
Fedorowicz, J., “Evolving Technology for Document-Based DSS,” in Sprague, R.,
Jr. and Watson, H. (eds.), Decision Support Systems: Putting Theory into Prac-
tice, 2nd ed. Englewood Cliffs, NJ: Prentice Hall, 1989.
Fjermestad, J., “An Analysis of Communication Mode in Group Support Systems
Research,” Decis Support Syst, 37(2), 2004.
Fjermestad, J. and S.R. Hiltz, “Group Support Systems: A Descriptive Evaluation
of Case and Field Studies,” J Manag Inform Syst, 17(3), 2000.
DSS Architecture and Types 187
Froelich, J., S. Ananyan and D.L. Olson, “Business Intelligence through Text
Mining,” Bus Intell J, 10(1), 2005.
George, J.F., J.F. Nunamaker and J.S. Valachich, “ODSS: Information Technol-
ogy for Organizational Change,” Decis Support Syst, 8(4), 1992.
Gray, D.A., “Airworthy: Decision Support for Aircraft Overhaul Maintenance
Planning,” ORMS Today, December, 1992.
Gray, P. and J.F. Nunamaker, “Group Decision Support Systems,” in Sprague, R.
and Watson, H. (eds.), Decision Support Systems: Putting Theory into Practice.
Upper Saddle River, NJ: Prentice-Hall, 1993.
Han, J. and M. Kamber, Data Mining: Concepts and Techniques. San Mateo, CA:
Morgan Kaufmann, 2000.
Hartono, E. and C.W. Holsapple, “Theoretical Foundations for Collaborative
Commerce Research and Practice,” Inform Syst e-Bus Manage, 2(1), 2004.
Holsapple, C.W., “Framework for a Generalized Intelligent Decision Support
System,” Ph.D. dissertation, Krannert Graduate School of Management, Purdue
University, 1977.
Holsapple, C.W., “The Knowledge System for a Generalized Problem Processor,”
Krannert Institute Paper, No. 827, Purdue University, 1983.
Holsapple. C.W., “Adapting Demons to Knowledge Management Environments,”
Decis Support Syst, 3(4), 1987.
Holsapple. C.W., “Knowledge Management in Decision Making and Decision
Support,” Knowl Policy, 8(1), 1995.
Holsapple, C.W. and K.D. Joshi, “An Investigation of Factors that Influence the
Management of Knowledge in Organizations,” J Strat Inf Syst, 9(2/3), 2000.
Holsapple, C.W. and K.D. Joshi, “Knowledge Manipulation Activities: Results of
a Delphi Study,” Inform Manage, 39(6), 2002.
Holsapple, C.W. and K.D. Joshi, “A Knowledge Management Ontology,” in
Holsapple, C.W. (ed.), Handbook on Knowledge Management, Volume 1. Ber-
lin: Springer, 2003.
Holsapple, C.W. and K.D. Joshi, “A Formal Knowledge Management Ontology:
Conduct, Activities, Resources, and Influences,” J Am Soc Inf Sci Tec, 55(7),
2004.
Holsapple, C.W., K.D. Joshi and M. Singh, “Decision Support Applications in
Electronic Commerce,” in Shaw, M., et
al. (eds.), Handbook on Electronic
Commerce. Berlin: Springer, 2000.
Holsapple, C.W. and M.P. Sena, “ERP Plans and Decision-Support Benefits,”
Decis Support Syst, 38(4), 2005.
188 Clyde W. Holsapple
Holsapple, C.W. and A.B. Whinston, “Software Tools for Knowledge Fusion,”
Computerworld, 17(15), 1983.
Holsapple, C.W. and A.B. Whinston, “Aspects of Integrated Software,” in Pro-
ceedings of the National Computer Conference, Las Vegas, July, 1984.
Holsapple, C.W. and A.B. Whinston, “Management Support through Artificial
Intelligence,” Hum Support Syst Manage, 5, 1985.
Holsapple, C.W. and A.B. Whinston, Manager’s Guide to Expert Systems.
Homewood, IL: Dow Jones-Irwin, 1986.
Holsapple, C.W. and A.B. Whinston, The Information Jungle: A Quasi-Novel
Approach to Managing Corporate Knowledge. Homewood, IL: Dow Jones-
Irwin, 1988.
Holsapple, C.W. and A.B. Whinston, Decision Support Systems: A Knowledge-
Based Approach. St. Paul: West, 1996.
Inmon, W.H., Building the Data Warehouse. New York: Wiley, 2002.
Joyce, J.D. and N.N. Oliver, “Impacts of a Relational Information System in In-
dustrial Decisions,” Database, 8(3), 1977.
Keen, P.G.W., “Decision Support Systems: The Next Decade,” Decis Support
Syst, 3(3), 1987.
Kersten G.E. and S.J. Noronha, “WWW-based Negotiation Support: Design, Im-
plementation, and Use,” Decis Support Syst, 25(2), 1999.
Klaas, R.L., “A DSS for Airline Management,” DataBase, 8(3), 1977.
Koutsoukis, N.S., G. Mitra and C. Lucas, “Adapting On-line Analytical Process-
ing for Decision Modelling: The Interaction of Information and Decision Tech-
nologies,” Decis Support Syst, 26(1), 1999.
Kwan, M.M. and P. Balasubramanian, “KnowledgeScope: Managing Knowledge
in Context,” Decis Support Syst, 35(4), 2003.
Lee, K. and S. Huh, “Model-Solver Integration in Decision Support Systems:
A Web Services Approach,” AIS SIGDSS Workshop, Seattle, December, 2003.
McGill, T.J. and J.E. Koblas, “The Role of Spreadsheet Knowledge in User-
Developed Application Success,” Decis Support Syst, 39(3), 2005.
Minch, R.P., “Application Research Areas for Hypertext in Decision Support
Systems,” J Manage Informn Syst, 6(2), 1989.
Nasukawa, T. and T. Nagano, “Text Analysis and Knowledge Mining System,”
IBM Syst J, 40(4), 2001.
Neundorf, K.A., The Content Analysis Guidebook. Thousand Oaks, CA: Sage,
2002.
DSS Architecture and Types 189
Nunamaker, J.F., D. Vogel, A. Heminger, B. Martz, R. Grohowski and C. Mcgoff,
“Experiences at IBM with Group Support Systems: A Field Study,” Decis Sup-
port Syst, 5(2), 1989.
Osborn, P.B. and W.H. Zickefoose, “Building an Expert System from the Ground
Up,” AI Expert, 5(5), 1990.
Rein G.L., C.W. Holsapple and A.B. Whinston, “Computer Support of Organiza-
tion Design and Learning,” J Org Comp, 3(1), 1993.
Santhanam, R., T. Guimaraes and J.F. George, An Empirical Investigation of
ODSS Impact on Individuals and Organizations,” Decis Support Syst, 30(1),
2000.
Sprague, R.H., Jr. and E. D. Carlson, Building Effective Decision Support Systems.
Englewood Cliffs, NJ: Prentice Hall, 1982.
Thierauf, R.J., User-oriented Decision Support Systems. Englewood Cliffs, NJ:
Prentice Hall, 1988.
Turban, E., Decision Support and Expert Systems. New York: Macmillan, 1988.
Turoff, M., “Computer-Mediated Communication Requirements for Group Sup-
port,” J Org Comp, 1(1), 1991.
... According to Keen and colleagues (1978), Decision Support Systems (DSSs) are IT-enabled tools that aim to enhance the effectiveness and efficiency of managerial and professional decision making for ill-structured problems. There exists a wide variety of Decision Support Systems, including passive DSSs that provide the user with compiled information only and active DSSs that provide specific solutions or recommendations (Holsapple, 2008). ...
... Finally, a Knowledge-driven DSS provides specialised problem-solving expertise stored as facts, rules, or procedures'. Holsapple (2008) divides the structural definition of Decision Support Systems into four essential components: ...
Thesis
Full-text available
Ecological Risk Assessment (ERA) is a process undertaken for estimating the environmental harms caused by human activities. The assessment is based on three components: effect assessment, exposure assessment, and risk characterisation. The latter is a combination of the former two. Various methodologies can be used for performing ERA, which can be categorised into deterministic and probabilistic. Probabilistic techniques have been at the focus of research the last years, due to their elaborated character and the possibilities they offer for more refined risk assessments. Despite their obvious advantages, probabilistic techniques present also disadvantages and challenges that need to be tackled. In the thesis, the possibility of exploring further the concept of Probabilistic Ecological Risk Assessment (PERA) is addressed. The main motivation of the thesis is identified in the effort to combine various well known concepts and methods for Ecological Risk Assessment, while enhancing them with new features and functionalities to serve the current needs of Risk Assessors. Therefore, providing a complete software package that allows performing efficient Propabilistic ERA (PERA) and offers related functionalities, all gathered in one place. The proposed software is developed as part of the research project AMORE (funded by the National French Research Academy – ANR). A proposal for a Decision Support System (DSS), named AMORE DSS, supporting Probabilistic ERA is described in detail and validated through the application of the proposed DSS to a case study for assessing the effects and risks posed by the presence of cyanide in a river in northwestern France. The AMORE DSS aims at allowing efficient Probabilistic ERA and tackles issues related with PERA and the concept of weighted Species Sensitivity Distributions (SSWD) such as the handling of uncertainty in PERA, the production of reliable SSWD graphs and the assessment of the quality of ecotoxicological data. The theoretical section of the thesis is split into two main parts. In the first, the concept of Ecological Risk Assessment is introduced and the principal methods of interest are described. It is followed by the presentation of the concepts of Multi-Criteria Decision Analysis (MCDA) and Decision Support Systems (DSS), which are important aspects of the developed research. The methodological developments of the thesis are based on a proposal for the estimation of the reliability and relevance of ecotoxicological data used in ERA, which is presented in detail and evaluated. The proposed methodology is based on Multi-Criteria Decision Analysis and allows the assessment of ecotoxicological data on the basis of a fixed set of criteria and mathematically stable and robust aggregation techniques. Therefore, the methodology suggests the production of reliable weighted Species Sensitivity Distributions, a vital component of the probabilistic ERA and the calculation of risk probabilities. The proposal allows incorporating in the risk assessment the knowledge gathered from an expert panel and gives significant strength to the risk assessors for the performed assessments, through the use of previously not widely available information and expertise. The proposed DSS is built on the three components (exposure, effects, risk) of ERA and provides a complete set of functionalities to the risk assessors, enhanced with unique features. The thesis describes in detail the development of the software and the functionalities of each of its modules. The Exposure Assessment module aims at providing to the Decision Maker/Risk Assessor a collection of tools for the statistical analysis of environmental exposure data, through the concept of Predicted Environmental Concentration. The Effect Assessment module is based on the concept of weighted Species Sensitivity Distributions (SSWD) and incorporates the proposed methodology for the assessment of the reliability and relevance of ecotoxicological data. The Risk Assessment module is based on the concept of Potentially Affected Fraction (PAF) and aims at synthesising the results of the previous two modules for the estimation of risks, in an efficient and easy to present way. The last part of the thesis is dedicated to the application of the DSS to a real life case study. A Risk Assessment process is performed for estimating the sensitivity of species to the presence of Cyanide (CN) in the environment, for estimating Environmental Quality Criteria (EQC) for the assessed case and for estimating the level of risk posed from Cyanide at the ecosystem. The assessed area is the Selune rivershed in the Manche department of the lower Normandy region in France, where four sampling stations have been identified with records of Cyanide presence for the period 2005-2014. Regarding the ecotoxicological data of the case study, 26 scientific articles on cyanide toxicity, published in the period 1965-2011, have been analysed for the extraction and assessment of 46 toxicological endpoints for the aquatic environment. The case study is firstly based on all the available ecotoxicological data and secondly based on data split per taxonomic groups (i.e vertebrates, invertebrates) and trophic levels (i.e. primary producers, primary/secondary consumers). Specifically, six (6) sets of SSWD graphs are produced (i.e. All data, Vertebrates, Invertebrates, Primary producers, Primary consumers, Secondary consumers) with the use of two weighting options: (i) the weighting coefficients that are produced with the application of the MCDA based methodology and (ii) equal weighting coefficients for all the data. A comprehensive comparison of the two types of SSWD is performed and discussed in detail for the identification of the appropriateness of the fitting of the SSD curves to the experimental data. Hazardous concentrations (HCx) are estimated and presented for all the taxonomic groups and trophic levels. In addition, in combination with the results of the statistical analysis of the environmental exposure data, the risk is estimated for the assessed stations in the case study area. The results of the case study show that the primary producers are found to be the most sensitive trophic level while Invertebrates are more sensitive as a taxonomic group for low cyanide concentrations and Vertebrates are more sensitive for higher concentrations. Regarding the calculated risk indices, station 3 (L’Yvrande) of the Manche region is the area with the higher estimated risk. The performed application of the DSS in the cyanide case study demonstrates a complete probabilistic Ecological Risk Assessment process with the use of Species Sensitivity Distributions and the utilisation of Multi-Criteria Decision Analysis. The case study, alongside with the validation of the developed DSS, demonstrates the performance of the proposed MCDA-based WoE framework for the analysis of ecotoxicological data, based on three distinctive Lines of Evidence (Experimental Reliability, Statistical Reliability, Biological Relevance). The framework and the related MCDA methodology constitute an innovative development in the field of quantitative ecotoxicological data assessment frameworks. Furthermore, a robust performance of the DSS has been identified, which allows potential for adoption within the risk assessment research fields.The thesis is concluded with future considerations for the developed DSS, which could provide interesting functionalities and extensions of the capabilities of the software.
... Neu istdas klang bereits einleitend andie Diskussion um computerbasierte Unterstützung innerhalb der Sozialen Arbeit keinesfalls. Seit den 1980er Jahren gibt es erste Überlegungen, die in der Personalverwaltung einsetzbaren Expertensysteme auch für die professionelle Arbeit nutzbar zu machen (Schoech et al. 1985), um "cognitive, temporal, spatial and/or economic limits on the decision maker" aufzubrechen (Holsapple 2008). Während zu Beginn der Forschung zu DSSs in den 1970er Jahren noch von einer Neutralität der Technik und der passiven Einbindung der entscheidungstragenden Personen ausgegangen wurde (Keen und Morton 1978), änderte sich dies bereits innerhalb der ersten Dekade, indem nichttechnische, gestalterische und entscheidungstheoretische Aspekte stärkere Berücksichtigung im technik-theoretischen Diskurs fanden (Keen 1987). ...
Chapter
Ein Beitrag, der sich mit ethischen und professionsspezifischen Aspekten algorithmischer Entscheidungsunterstützungssysteme innerhalb der Teilhabeplanung auseinandersetzt, muss gleich zwei Herausforderungen meistern: Einerseits gilt es, die spezifischen Besonderheiten algorithmischer Systeme im Blick zu behalten und die potenziellen Herausforderungen dieser Technik im Rahmen der Eingliederungshilfe herauszuarbeiten. Andererseits müssen die professionsspezifischen Herausforderungen dargelegt werden, die sich durch die potenzielle Implementierung algorithmischer Systeme ergeben. Der vorliegende Beitrag kann daher nicht als umfassende oder gar abschließende Auseinandersetzung mit dem Thema, sondern nur als Aufschlag für eine anstehende, interdisziplinär ausgerichtete Diskussion verstanden werden. - - - - 1. Einleitung 2. Verortung in der aktuellen Diskurslandschaft 2.1. Informations- und Kommunikationstechnologie in der Sozialen Arbeit 2.2. Digitalisierung in der öffentlichen Verwaltung 2.3. Änderungen in der Eingliederungs- und Behindertenhilfe 3. Herausforderungen 3.1. Bias und fehlende Nachvollziehbarkeit 3.2. Skepsis gegenüber Standardisierung und Steuerungsbestrebungen 4. Fazit Förderhinweis Quellen
... While programming language is considered as the technical focus of a language system, a presentation system refers to the user interface. Moreover, a knowledge system includes any input such as data, document and datasets [74] Starting with the language system, studies in the healthcare industry present various programming languages such as C #, Visual Basic, and Phyton and database management systems such as MySQL. Dios et al. [20] presented a linear programming model for hospital management as an administrative decision support application. ...
Article
Full-text available
Healthcare professionals and inter (or multi) disciplinary academia have been paying more attention to decision support systems ( for improved decision making during their health service processes or management, as well as clinical practices. Although there have been numerous DSS applications in the healthcare industry, it has been intended to provide a categorical snapshot view of current implementations or academic work at specific DSS types for better understanding the application domains by addressing the gap in the literature. To achieve this, it has been focused on DSS applications in healthcare specifically by concentrating on two main types: model driven and knowledge driven. In this context, re levant information systems and medical science literatures were reviewed. For health service problems like hospital placement decisions and homecare route planning, model driven DSS applications are used for optimization and modelling. Both conventional op erations research techniques like optimization, decision analysis, simulation, and multi criteria decision making, as well as contemporary ones like heuristic search, benefit from these applications. In addition, artificial intelligence techniques help hea lth decision makers via knowledge driven DSS applications, specifically clinical decision support systems ( Artificial intelligence applications can also assist health professionals in enhancing their decision making abilities by incorporating comple x operational rules and developing such procedures as single or multi agent systems. This research focuses on what to emphasis on while designing a DSS in the healthcare setting, such as which programming or modelling languages to employ and how to transform a model driven DSS into a knowledge driven DSS, or how to create the DSS more intelligent. Overall, this study indicates a present course for DSS and offers useful knowledge for both scholars and professionals in the healthcare domain.
... However, DSS artifacts are often criticized for their lack of relevance to practice and for neglecting configurability and contextual dynamics [19][20]. To counteract this lack, we oriented our DSS prototype development towards essential DSS design elements [21]. We worked towards the highest possible practical relevance, configurability and contextual dynamics [19]. ...
Chapter
Global urbanization for decades has led to unprecedented levels and growing demands for urban logistics. Thus, problems such as congestion, environmental noise, and urban sprawl are growing. As a result, many cities face problems of optimal decision-making regarding green and sustainable smart transportation systems and infrastructures. However, various possible measures and logistics concepts are available to improve urban logistics, while effects are unclear and difficult to predict. To meet the growing need for future-oriented decisions by city authorities, we developed a decision support system prototype that allows a strategic simulation-based evaluation of different logistics concepts regarding defined targets, e.g., pollutant emissions, traffic flow, space requirements, or economic efficiency on a city district level. An expert system for the strategic evaluation of logistics concepts on a city district level is integrated to achieve transferability and scalability.
... However, DSS artifacts are often criticized for their lack of relevance to practice and for neglecting configurability and contextual dynamics [19][20]. To counteract this lack, we oriented our DSS prototype development towards essential DSS design elements [21]. We worked towards the highest possible practical relevance, configurability and contextual dynamics [19]. ...
Conference Paper
Global urbanization for decades has led to unprecedented levels and growing demands for urban logistics. Thus, problems such as congestion, environmental noise, and urban sprawl are growing. As a result, many cities face problems of optimal decision-making regarding green and sustainable smart transportation systems and infrastructures. However, various possible measures and logistics concepts are available to improve urban logistics, while effects are unclear and difficult to predict. To meet the growing need for future-oriented decisions by city authorities, we developed a decision support system prototype that allows a strategic simulation-based evaluation of different logistics concepts regarding defined targets, e.g., pollutant emissions, traffic flow, space requirements, or economic efficiency on a city district level. An expert system for the strategic evaluation of logistics concepts on a city district level is integrated to achieve transferability and scalability.
... In knowledge-based systems, models that are rule or suggestion-oriented are more prevalent. In model-driven systems the majority come from representational, excel-based, solver-based, and optimization models (Hasan et al. 2017;Holsapple et al. 2008;Power 2004;Alter 1980). Figure 1 highlights the taxonomies around AI, DSSs, and OR. ...
Article
Full-text available
Operations research (OR) has been at the core of decision making since World War II, and today, business interactions on different platforms have changed business dynamics, introducing a high degree of uncertainty. To have a sustainable vision of their business, firms need to have a suitable decision-making process at each stage, including minute details. Our study reviews and investigates the existing research in the field of decision support systems (DSSs) and how artificial intelligence (AI) capabilities have been integrated into OR. The findings of our review show how AI has contributed to decision making in the operations research field. This review presents synergies, differences, and overlaps in AI, DSSs, and OR. Furthermore, a clarification of the literature based on the approaches adopted to develop the DSS is presented along with the underlying theories. The classification has been primarily divided into two categories, i.e. theory building and application-based approaches, along with taxonomies based on the AI, DSS, and OR areas. In this review, past studies were calibrated according to prognostic capability, exploitation of large data sets, number of factors considered, development of learning capability, and validation in the decision-making framework. This paper presents gaps and future research opportunities concerning prediction and learning, decision making and optimization in view of intelligent decision making in today’s era of uncertainty. The theoretical and managerial implications are set forth in the discussion section justifying the research questions.
... DSS designers thus need to move beyond traditional DSS architectures featuring the interrelationships of the major essential DSS design elements (Holsapple, 2008) as their source of knowledge about design. Even DSS generators, generally based on traditional DSS architectures within a package of related integrated software that provides a set of capabilities to quickly and easily develop a specific DSS (Power, 2002), are likewise of limited use for designers keen to overcome the lack of relevance challenges associated with DSS. ...
... Based on Power (2008) and Holsapple (2008) a taxonomy of decision support systems (DSS) is proposed, following efforts of authors in other works (Marques et al., 2018), which covers the various dimensions and requirements mentioned above. In a highly simplified way, three types or levels can be assumed, depending on the information they use and the type of answers that can be obtained: i) situational (diagnostic); ii) forecast (projection); and iii) interactive (foresight). ...
Chapter
Full-text available
Housing is more than a living space; it represents a social status capable of promoting several forms and levels of socialization and segregation. Housing is characterized by a set of attributes and functions that are valued differently, in which the consumption of “good housing” is made considering the preferences of their attributes, perceived tangibly/intangibly by the consumers. The reciprocal relation between housing and space, and the lack of structured information capable of apprising the key drivers of such dimensions increase the complexity of understanding the rational mechanisms of evaluating the real housing value. Despite the challenges associated with the modelling of housing markets, urban studies and spatial econometric literature provide a broad (but unfinished) theoretical framework and practical tools that are able to describe, understand, and predict households' housing consumption. Thus, the aim of this chapter is to present concepts and techniques to rationally capture individual and collective housing preferences, and the way in which they interact.
Article
While human-centered system design has actively sought to reduce cognitive biases in decision-making, a gap exists on the distinct role of media features in decision-support systems (DSS). Intuitively, we may presume that DSS interactivity and revisability remedy such biases. Yet, experiments reported herein reveal that combining the two features may conversely aggravate such bias. These features have been studied as distinct factors moderating bias in decision-criteria rating tasks in two experiments (sample sizes: n1 =96 for first; for second with criteria vividness controlled, n2=429). Such study emanates from a novel view of DSS as persuasive communication media. Thus, a unique adaptation of Media Synchronicity Theory to a computer–human communication context ensues. This promotes the isolated study of interactivity and revisability, alongside their interaction. Hence, the study finds statistically significant bias reductions for revisability treatments without interactive feedback. Such real-time automatic feedback, surprisingly, makes no impact in the majority of cases. Vividness controlled, it is still revisability-only cases indicating any reduction in possible order effects bias (despite extra effort required). Thus, a fresh case is made to evaluate interactivity and revisability separately in DSS media, with empirical evidence that – together – they could be ‘doing more harm than good’.
Article
Full-text available
Virtual coaching has emerged as a promising solution to extend independent living for older adults. A virtual coach system is an always-attentive personalized system that continuously monitors user’s activity and surroundings and delivers interventions – that is, intentional messages – in the appropriate moment. This article presents a survey of different approaches in virtual coaching for older adults, from the less technically supported tools to the latest developments and future avenues for research. It focuses on the technical aspects, especially on software architectures, user interaction and coaching personalization. Nevertheless, some aspects from the fields of personality/social psychology are also presented in the context of coaching strategies. Coaching is considered holistically, including matters such as physical and cognitive training, nutrition, social interaction and mood.
Chapter
Knowledge-based organizations are hosts for multitudes of knowledge management episodes. Each episode is triggered by a knowledge need (or opportunity) and culminates with the satisfaction of that need (or its abandonment). Within an episode, one or more of the organization’s processors (human and/or computer-based) manipulate knowledge resources in various ways and subject to various influences in an effort to meet the need or take advantage of the opportunity. This chapter presents an extensive ontology of knowledge management. The ontology identifies and characterizes basic components of knowledge management episodes, the knowledge resources an organization uses in these episodes, a generic set of elemental knowledge manipulation activities that manifest within knowledge management episodes, and categories of influences on the conduct and outcomes of these episodes. This ontology was developed using conceptual synthesis and a collaborative methodology involving an international panel of researchers and practitioners in the knowledge management field. The ontology can serve as a common language for discourse about knowledge management. For researchers, it suggests issues that deserve investigation and concepts that must be considered in explorations of knowledge management episodes. For practitioners, the ontology provides a perspective on factors that need to be considered in the implementation of an organization’s knowledge management initiatives.
Article
The value and nature of a mixed approach to knowledge representation is examined. The mixed approach utilizes data base techniques and expressions in first-order predicate calculus. The nature of a general problem processor that can use information organized according to the mixed approach is discussed and illustrated. The user's language interface with this problem processor is non-procedural and English-like. The utilization of predicate calculus axioms for data integrity and program module management is also explored.
Article
The interrelationships between organization structure and information systems have attracted the attention of researchers since the 1950s. We propose three different architectures for organizational decision support systems (ODSS), each of which is tailored to an emerging change in organization structures: Downsizing; a focus on teams; and outsourcing. We argue that no single ODSS architecture can adequately meet the challenges of each structural change. Rather, each proposed architecture relies to a lesser or greater degree on five types of information technology: Communication; co-ordination; filtering; decision making; and monitoring technologies.