What are the features of a knowledge-driven DSS?

by Dan Power


This is installment #4 about features of computerized decision support systems. The focus is on knowledge-driven Decision Support Systems (KDSS). The goal is to identify major, observable aspects, attributes or elements from a user's perspective that distinguish knowledge-driven DSS from other types of DSS (Power, 2002) and from other computerized information systems.

In general, a knowledge-driven DSS suggests or recommends actions to targeted users. This type of DSS has specialized problem-solving expertise relevant to a specific narrow task. The "expertise" consists of knowledge about a particular problem domain, understanding of problems within that domain, and "skill" at solving one or some of these problems. The system has subject-specific knowledge of one or more human experts. These systems have also been called advisory systems, consultation systems, suggestion systems (Alter, 1980), knowledge-based systems (Klein and Methlie, 1995), recommender systems, rule-based DSS, and management expert systems.

Knowledge-driven DSS can store and apply knowledge for a variety of specific problems/tasks that would otherwise be resolved by a human expert. The generic tasks include classification, configuration, diagnosis, interpretation, planning and prediction. Historically, diagnosis has been the most popular DSS application area. Building a specific knowledge-driven DSS depends upon the demand for the system and the anticipated benefits. As with all DSS, the goal is supporting a human decision maker in completing the task, rather than replacing the decision maker. The programming and development tools used to build these systems are from Artificial Intelligence and Statistics. The systems may be rule-based, statistics-based, heuristic, object-based, logic-based, or induction based. Some systems use more than one technology.

Classification involves separating a specific instance into a broader class based upon characteristics. Configuration involves creating an arrangement of objects given performance criteria or constraints. Diagnosis involves hypothesizing a cause given symptom and situational information. Interpretation refers to adding meaning, explanation and possibly understanding in a specific situation or context. Planning usually involves sequencing an assortment of actions or means to achieve desired ends in a constrained situation. Finally, prediction refers to identifying and forecasting a future state of a system.

For many years, I have been interested in building this type of DSS (cf., Power, 1985). The major barrier to making progress has been associated with deployment of the DSS to make the knowledge readily available more so than with resistance to the use of such systems. In recent years, Web technologies and hand-held and tablet PCs have made deployment of knowledge-driven DSS much easier and much less expensive.

Case study examples of knowledge-driven DSS at include Huntington (2006), Biss (2002), Exsys Staff (2002) and Pontz and Power (2002). Huntington, CEO at Exsys, explains how the U.S. Small Business Administration is using his company's product to share knowledge using the Web. Biss reports a prototype medical decision support system built using Bayesian technology from Hugin ( The EXSYS Staff case reports a project to support families in making decisions about overseas staff assignments. The Pontz and Power case reports building a knowledge-driven DSS called EASE at the Pennsylvania Department of Labor and Industry.

Two recent press releases at document a number of significant developments. On October 10, 2006, Exsys announced it was adapting its CorvidŽ software to run on HP iPAQ pocket PCs and other personal digital assistants (PDAs). The "new capability enables companies to capture and provide decision support expertise on handheld mobile devices in a wide variety of areas including diagnostics, sales support and regulatory compliance." On January 10, 2006, Epocrates Inc., provider of mobile and desktop clinical applications, announced the launch of the Epocrates SxDx(TM) disease diagnosis and treatment reference and symptom assessment tool. The application, developed in collaboration with Massachusetts General Hospital's Laboratory of Computer Science, "provides intelligent decision support to clinicians throughout the diagnosis process. Developed for mobile devices, including personal digital assistants, the new product allows healthcare professionals to enter patient symptoms and findings to generate a clinically useful diagnosis index to the Epocrates disease reference."

The following are major features of knowledge-driven DSS from a user's perspective:

1) Asks questions. Historically knowledge-driven DSS attempt to create an interactive dialogue with users that simulates an interrogation by a "real" expert. A key feature is interactivity with the user and contingent branching based upon responses. In some systems, the initial questions may be implied on an input form.

2) Backtrack capability. Users can often move backward through the questions and alter responses. This feature makes it easy to go back over the course by which one reached a result and then change a subjective judgment and change a recommendation/result. In some systems it may be useful to record that backtracking occurred.

3) Display confidence or certainty information. Numeric values called confidences, likelihoods or ranks can be calculated in some systems. A confidence interval is a statistical range with a specified probability that a given result lies within the range. The DSS may be able to create a confidence interval for a recommendation or diagnosis. When these capabilities are present in the design and development environment, a user can display the confidence information.

4) Explain HOW. After a knowledge-driven DSS has reached a solution or conclusion for the problem, the user can often request an explanation of how the solution/conclusion was reached. This is one of the most powerful features and is commonly present in a knowledge-driven DSS. This feature can enhance user confidence in the recommendation and hence acceptance of the system.

5) Explain WHY. Users should be able to ask why the system is asking a specific question. This feature helps explain the process to the user and enhances the user's expertise.

6) Initiate actions. In some knowledge-driven DSS, users can send an email and or otherwise implement a recommendation. A good example of this feature is in Pontz and Power (2002). Claims examiners using EASE can send unemployment claims determinations directly to a centrally located printer where the determination is inserted automatically into an envelope with the appropriate postage.

7) Output selection. In addition to displaying confidence information, some knowledge-driven DSS have multiple formats for displaying outputs. For example, it may be possible to display a diagram of a configuration or a graphical depiction of a plan.

8) Resume analysis. In some situations it may be desirable to pause during an analysis and resume the analysis at a later point in time. The system needs to display active and incomplete analyses.

9) Retrieve data about a specific case or instance. Data used in some knowledge-driven DSS may come from other computerized sources and the KDSS user must be able to retrieve data from external sources. For example, a diagnostic system may retrieve information about a patient from laboratory tests or a configuration system may need data from an inventory system. When this feature is implemented, users also need access to data source field definitions.

10) Store inputs, results and user actions. This may be a discretionary or mandatory feature of the system. If the feature is mandatory the user should be aware that data is being captured by the system for monitoring and control purposes.

11) Train users. Some KDSS have sample or test cases to help train users for future use of the system in actual situations. The KDSS would then have a feature to start a training or tutoring mode.

Please note: Decisions made using knowledge-driven DSS can be biased by the responses provided by the user. Potentially a KDSS can reduce the risk of human errors in well-structured situations where experts perform better than novices, but novice users must avoid distorting or biasing responses to KDSS questions. Decision making is a cognitive process and knowledge-driven DSS attempt to support and enhance a person's reasoning and thinking in a specific narrow domain. If novice users begin using a KDSS with a preconceived notion or conclusion, the confirmation bias phenomenon will often result in biased responses to confirm the prior decision, conclusion or hypothesis.

The prospects and benefits for managing knowledge and supporting decision making using knowledge-driven DSS is evolving. KDSS can increase the distribution of expertise; broaden job descriptions for individual workers; and create a new communication channel for knowledge. As Pontz and Power (2002) suggest, using a KDSS can result in more consistent decisions and can create efficiencies and reduce the time needed to solve problems. KDSS can also reduced training costs and rapidly disseminate and update large amounts of knowledge. Finally, KDSS can help centralize control of repetitive, structured decision making processes.

As always your comments, questions and suggestions are welcomed.


Alter, S.L. Decision Support Systems: Current Practice and Continuing Challenge. Reading, MA: Addison-Wesley, 1980.

Biss, Andrew, "Dynasty Triage Advisor Enables Medical Decision Support", 2002, posted at DSSResources.COM December 21, 2002.

EXSYS Staff, "IAP Systems Using Exsys CORVID Expert System Software to Support Corporate Families on Overseas Assignments", EXSYS, Inc., 2002, posted at DSSResources.COM May 12, 2002.

Georgia Tech Research Institute Staff, "Assistance to the Emergency Response Community", Georgia Tech Research Institute, 2001.

Huntington, D., "From Information to Answers: Transferring Expertise at the SBA", posted at, November 3, 2006.

Klein, M. and L. B. Methlie, Knowledge-based Decision Support Systems with Applications in Business. Chichester, UK: John Wiley & Sons, 1995.v Pontz, C. and D. J. Power, "Building an Expert Assistance System for Examiners (EASE) at the Pennsylvania Department of Labor and Industry", November 2002, posted at DSSResources.COM November 14, 2002.

Power, D.J., Using the Symptoms, Problems and Treatment Framework to Structure Knowledge for Management Expert Systems, In: Methlie, L. and Sprague, R. (Ed.), Knowledge Representation for Decision Support Systems, New York: North-Holland Publishing Co., 1985, pps. 245-254.

Power, D. Decision Support Systems: Concepts and Resources for Managers, Westport, CT: Quorum/Greenwood, 2002.

Press Releases

Epocrates, "Epocrates collaborates with Massachusetts General Hospital to offer unique decision support tool,", 01/10/2006.

EXSYS, "On-the-go employees and customers can now access expert corporate advice via handheld mobile devices,", 10/10/2006.

Web links expert-systems.htm

Last update: 2008-08-09 11:33
Author: Daniel Power

Print this record Print this record
Show this as PDF file Show this as PDF file

Please rate this entry:

Average rating: 1.54 from 5 (72 Votes )

completely useless 1 2 3 4 5 most valuable

You cannot comment on this entry

DSS Home |  About Us |  Contact Us |  Site Index |  Subscribe | What's New
Please Tell Your Friends about DSSResources.COMCopyright © 1995-2015 by D. J. Power (see his home page).
DSSResources.COMsm is maintained by Daniel J. Power. Please contact him at with questions. See disclaimer and privacy statement.


powered by phpMyFAQ 1.5.3