DSS Case Summaries

This catalog currently focuses on "classic" published examples of Decision Support Systems from the academic literature and from a limited number of vendor web sites. Most of the DSS are early prototype systems. I hope you will find something useful in this list.


The following are specific Decision Support Systems, that is the DSS were implemented to accomplish a specific decision support task.

  • AAIMS, An Analytic Information Management System
  • Advanced Scout
  • BCA DSS, Base Closure and Analysis DSS
  • CARGuide, Computer for Automobile Route Guidance
  • CATD
  • DELTA, Dieasel-Electric Locomotive Troubleshooting Aid
  • GADS, Geodata Analysis and Display System
  • GCA, Graduate Course Adviser
  • Hannaford DSS Web
  • ISIS, Intelligent Scheduling and Information System
  • KNOBS, KNOwledge Based System
  • PMS, Portfolio Management System
  • RealPlan
  • XCON

    AAIMS This system was implemented by American Airlines in mid the 1970s. It was developed in APL and it was used for data analysis and report generation. [Klass, R. L. A DSS for Airline Management. Data Base (Winter), pp. 3-8, 1977. Taplin, J. M. AAIMS: American Airlines Answers the What-Ifs. Infosystems (February, pp. 40-41, 1973.]

    Advanced Scout IBM has prototyped leading edge technology, called data mining, to help National Basketball Association (NBA) coaches and league officials organize and interpret the data collected at every game. Using data mining software called Advanced Scout to prepare for a game, a coach can quickly review countless stats: shots attempted, shots blocked, assists made, personal fouls. But Advanced Scout can also detect patterns in these statistics that a coach may not have known about. So during a game, a coach can know exactly which plays are most effective with which players and under what circumstances. Advanced Scout software provides an easier and more meaningful way to process information. "It helps coaches easily mine through and analyze a lot of data and no computer training or data analysis background is required," says Dr. Inderpal Bhandari, computer scientist at IBM's T.J. Watson Research Center in Hawthorne, N.Y. Patterns found through data mining are linked to the video of the game. This lets a coach look at the just those video clips that make up the interesting pattern. Incorporating this information into the strategy for a game can help the team win. Check www.research.ibm.com/scout/. See architecture diagram.

    AIRPLAN assisted air operations officers with the launch and recovery of aircraft on a carrier. The system analyzed current information (e.g., the aircraft's fuel level, the weather conditions at a possible divert site) and alerted the air operations officer of possible impending problems. AIRPLAN assessed the seriousness of a situation and managed its use of time by attending first to the most significant aspects of a problem. AIRPLAN was a rule-based system imlemented in OPS7. The system was developed at Carnegie-Mellon University. It reached the stage of a field prototype. [Masui, S., McDermott, J., and Sobel, A. Decision-making in time-critical situations. Proceedings IJCAI-83, pp. 233-235, 1983. See also Waterman, D. A. A Guide to Expert Systems. Reading, MA.: Addison-Wesley Pub. Co., pp. 237-379]

    ANALYST assisted field commanders with battlefield situation assessment. The system generated displays of enemy combat unit deployment and did so in real time from multisource sensor returns. ANALYST aggregated information from these multiple sensor sources to 1) locate and classify enemy battlefield units by echelon, general function, and relative location, and 2) detect force movement. The system contained expertise obtained from intelligence analysts, including how to interpret and integrate sensor data. ANALYST was implemented in FRANZ LISP and represented knowledge using a combination of frames and rules. It was a prototype developed at the MITRE Corporation. [1. Bonasso, R.P. Expert systems for intelligence fusion. Proceedings of the Army conference on Applications of AI to Battlefield Information Management, Report AD-A139 685, Battelle Columbus Laboratories, Washington, D.C., April 1983., 2. Bonasso, Jr., R.P. ANALYST, an expert system for processing sensor returns. Report MTP-83W00002, MITRE Corporation, 1820 Dolly Madison Blvd., McLean, Va., Febrary 1984. See also Waterman, D. A. A Guide to Expert Systems. Reading, MA.: Addison-Wesley Pub. Co., pp. 237-379]

    BCA DSS (Base Closure and Analysis DSS) According to a detailed Microstrategy case study, the BCA DSS provided the U.S. Air Force with a robust methodology and common framework for analyzing the impact of various base closure scenarios. A multi-layer, hierarchical filtering process is used to evaluate the relative impact of closing each base. Bases which pose minimum strategic, operational, social, and economic impact are placed at the top of the closure recommendation list. At any step, base closing committee members can review DSS-developed impact analyses to assist in determining which bases should proceed to the next level of analysis. Using the DSS, the BCEG members can perform analyses using the eight main criteria and 212 sub-criteria on which all bases are evaluated. These criteria, specified by DOD, focus on elements that impact operational effectiveness, including such items as alternate airfield availability, weather data, and facility infrastructure capacity. (from URL http://www.strategy.com/success/msi_saf1.htm) Check local archived copy.

    BRANDAID This DSS was developed by John D. C. Little and it was used for marketing mix analysis (price, promotion, place, and product). The system was implemented in EXPRESS (Information Resources, Inc.). [1. Little, John D. C. BRANDAID: A Marketing-Mix Model, Part 1: Structure. Operations Research 23, no. 4, (July-August), pp. 628-55, 1975. 2. Little, John D. C. BRANDAID: A Marketing-Mix Model, Part 2: Implementation, Calibration and Case Study. Operations Research 23, no. 4, (July-August), pp. 656-73, 1975.]

    CARGuide (Computer for Automobile Route Guidance) was designed to help drivers find routes and navigate in city streets. The system uses the starting and destination locations, together with its map information, to calculate an optimum route from starting point to destination. Optimum route finding is accomplished using a combination of a divide-and-conquer method, precomputed routes, and Dijkstra's shortest-path algorithm. Once found, the route is displayed and highlighted on a graphical display of a street map. During the trip the car's position along the route is updated and displayed. Before each intersection, the system pronounces a direction (straight, left, or right) and the name of the street to take. CARGuide was a demonstration prototype developed at Carnegie-Mellon University. [1. Sugie, M., Menzilcioglu, O., and Kung, H.T. CARGuide--onboard computer for automobile route guidance. Proceedings of the National Computer Conference, 1984. See also Waterman, D. A. A Guide to Expert Systems. Reading, MA.: Addison-Wesley Pub. Co., pp. 237-379]

    CATD or Computer Aided Train Dispatching was developed by the Southern Railway Co. from 1975 to 1982. It was initially built as a mini-computer based simulator and was installed and tested on the North Alabama track system in January 1980. The system was placed in production for that system on September 15, 1980. Gradually additional track systems were converted to CATD. The system provides decision support to aid train dispatchers in centralized traffic control. The system is linked to an automatic on station reporting signal and recordkeeping system. The system significantly reduced delays and reduced train meetings in the system. Over all delay per meet statistics were improved 15.5% during the 80 weeks following initial implementation. This translates into a cost saving of $316,600 per year for just the North Alabama track system. Full-scale implementation was expected to generate $3 million in annual cost savings. [Sauder, R. L. and Westerman, W.M. Computer Aided Train Dispatching: Decision Support Through Optimization. Interfaces, Vol. 13, No. 6, December 1983.]

    DELTA (Dieasel-Electric Locomotive Troubleshooting Aid) helps maintenance personnel to identify and correct malfunctions in diesel electric locomotives by applying diagnostic strategies for locomotive maintenance. The system can lead the user through a repair procedure. It is a rule-based system developed in a general-purpose representation language written in LISP. DELTA accesses its rules through both forward and backward chaining and uses certainty factors to handle uncertain rule premises. Although the system was prototyped in LISP, it was later reimplemented in FORTH for installation on microprocessor-based systems. The General Electric Company developed this system at their research and development center in Schenectady, New York. Current status unknown. It was field tested. [1. Marcus, Steven J. Computer systems applying expertise. The New York Times, August 29, 1983., 2. Bonissone, P. P. and Johnson, H. E. Expert system for diesel electric locomotive repair. Knowledge-based Systems Report, General Electric Co., Schenectady, N.Y., 1983., 3. DELTA/CATS-1. The Artificial Intelligence Report, vol.1, no.1, January 1984. See also Waterman, D. A. A Guide to Expert Systems. Reading, MA.: Addison-Wesley Pub. Co., pp. 237-379]

    DIPMETER ADVISOR reached conclusions about subsurface geological structure by interpreting dipmeter logs, measurements of the conductivity of rock in and around a borehole as related to depth below the surface. The system used knowledge about dipmeter patterns and geology to recognize features in the dipmeter data and relate them to underground geological stucture. The system provided the user with a menu-driven graphical interface and used a rule-base controlled by forward chaining. It was implemented in INTER-LISP-D and operated on the Xerox 1100 series workstations. The system was a research prototype developed by Schlumberger-Doll Research. Current status unknown. [1. Austin, H. Market trends in artificial intelligence. In W. Reitman (ed.) Artificial Intelligence Applications for Business, Norwood, N.J.: Ablex, 1984., 2. Smith, R. G. On the development of commercial expert systems. The AI Magazine, vol.5, no.3, Fall 1984., 3. Smith, R.G. and Young R.L. The design of the DIPMETER ADVISOR system. ACM Conference Proceedings, November 1984.]

    FOLIO helped portfolio managers determine client investment goals and select portfolios that best meet those goals. The system determined the client's needs during an interview and then recommends percentages of each fund that provide an optimum fit to the client's goals. FOLIO recognized a small number of classes of securities (e.g., dividend-oriented, lower-risk stocks and commodity-sensitive, higher-risk stocks) and maintained aggregate knowledge about the properties (e.g., rate of return) of the securities in each class. The system used a forward chaining, rule-based representation scheme to infer client goals and a linear programming scheme to maximize the fit between the goals and the portfolio. FOLIO was implemented in MRS. This demonstration prototype was developed at Stanford University. [1. Cohen, P. and Lieberman, M.D. A report on FOLIO: An expert assistant for portfolio managers. Proceedings IJCAI-83, pp. 212-214, 1983. See also Waterman, D. A. A Guide to Expert Systems. Reading, MA.: Addison-Wesley Pub. Co., pp. 237-379]

    GADS was an interactive system also known as Geodata Analysis and Display System. The goal in developing GADS was to enable nonprogrammers to solve unstructured problems more effectively by applying their job-specific experience and their own heuristics. It had a strong graphic display and "user-friendly" characteristics that enabled non-computer users to access, display, and analyze data that have geographic content and meaning. The system was used initially by police officers to analyze data on "calls for service". By 1982 17 specific DSS had been developed using GADS. [1. Sprague, R.H. and Carlson, E.D. Building Effective Decision Support Systems. Englewood Cliffs, NJ: Prentice-Hall, 1982.]

    GCA (Graduate Course Adviser) was designed to help graduate students plan their computer science curriculum. The system gathered information about a student's academic history and interests and then acted as a faculty adviser by suggesting a schedule of courses for the student. GCA's expertise included departmental and university regulations regarding graduate degree programs, course descriptions, and sequences of courses frequently taken by computer science students. The knowledge in GCA was organized as four interacting subsystems under the direction of a manager program. These subsystems determined 1) the number of courses the student should take, 2) the courses the student is permitted to take, 3) the best courses to take, and 4) the best schedule for the student. GCA's knowledge was encoded as rules with associated certainty factors. The system was implemented in PROLOG using a MYCIN-like inference engine. It was developed at Duke University and reached the stage of a research prototype. [1. Valtorta, M.G., Smith, B.T., and Loveland, D.W. The graduate course advisor: a multi-phase rule-based expert system. Proceedings of the IEEE Workshop on Principles of Knowledge-Based Systems, IEEE Computer Society, IEEE Computer Society Press, 1109 Spring Street, Silver Spring, Md., 1984.]

    Hannaford DSS Web Hannaford Brothers Grocery chain developed a DSS using Microstrategy's DSS Web. At Hannaford, DSS Web provides store managers with access to the same data warehouse application relied upon by corporate decision makers. Utilizing DSS Web, managers receive detailed sales, cost, inventory, and budget reports and use this information to make decisions at the store level. The February 1996 case study was originally at http://www.strategy.com/success/hanna.htm. Check local archived copy.

    ISIS (Intelligent Scheduling and Information System) constructed factory job shop schedules. The system selected a sequence of operations needed to complete an order, determined start and end times, and assigned resources to each operation. It also acted as an intelligent assistant, using its expertise to help plant schedulers maintain schedule consistency and identify decisions that resulted in unsatisfied constraints. Knowledge in the system included organizational goals such as due dates and costs, physical constraints such as limitations of particular machines, and causal constraints such as the order in which operations must be performed. ISIS used a frame-based knowledge representation scheme together with rules for resolving conflicting constraints. It was developed at Carnegie-Mellon University and tested in the context of a Westinghouse Electric Corporation turbine component plant. [1. Fox, M.S. and Smith, S.F. ISIS: a knowledge-based system for factory scheduling. Expert Systems, vol.1, no.1, 1984.]

    KNOBS (KNOwledge Based System) helped a controller at a tactical air command and control center perform mission planning. The system used knowledge about targets, resources, and planned missions to check the consistency of plan components, to rank possible plans, and to help generate new plans. Knowledge in KNOBS was in the form of frames and backward chaining rules, and it used a natural language subsystem for data base queries and updates. In the KNOBS literature, early articles refer to KNOBS as the expert system for mission planning. Later articles use the term KNOBS to mean the KNOBS architecture rather than a specific expert system. The system was implemented in FRL and ZETALISP. It was developed by the MITRE Corporation and reached the stage of a research prototype. [1. Scarl, E.A., Engelman, C., Pazzani, M.J., and Millen, J. The KNOBS system. MITRE Report, MITRE Corporation, 1983., 2. Kashner, F. Artificial intelligence aids military planners. Electronic business, pp. 155-156, December 1983. See also Waterman, D. A. A Guide to Expert Systems. Reading, MA.: Addison-Wesley Pub. Co., pp. 237-379]

    MYCIN assisted physicians in the selection of appropriate antimicrobial therapy for hospital patients with bacteremia, meningitis, and cystitis infections. The system diagnosed the cause of the infection using knowledge relating infecting organisms with patient history, symptoms, and laboratory test results. The system recommended drug treatment (type and dosage) according to procedures followed by physicians experienced in infectious disease therapy. MYCIN was a rule-based system employing a backward chaining control scheme. It included mechanisms for performing certainty calculations and providing explanations of the system's reasoning process. MYCIN was implemented in LISP. It was developed at Stanford University and reached the stage of a research prototype. [1. Buchanan, B. and Shortliffe, E. The problem of evaluation. In Buchanan and Shortliffe (eds.) Rule-Based Expert Systems, Reading, Mass.: Addison-Wesley, pp. 571-596, 1984., 2. Buchanan, B. and Shortliffe, E. Uncertainty and evidential support. In Buchanan and Shortliffe (eds.) Rule-Based Expert Systems, Reading, Mass: Addison-Wesley, pp. 209-232, 1984., 3. Buchanan, B. and Shortliffe, E. Use of MYCIN inference engine. In Buchanan and Shortliffe (eds.) Rule-Based Expert Systems, Reading, Mass.: Addison-Wesley, pp. 295-301, 1984.]

    PMS T. P. Gerrity designed the Portfolio Management System and it was implemented in four banks beginning in 1974. The purpose of the DSS was to help manage security portfolios and manage risk and return. The DSS included commands like STATUS to display the contents of a portfolio; TABLE to display portfolio values; and SCATTER and HIST to display scatter plots and histograms. [Keen, P.G.W. and M. S. Scott-Morton. Decision Support Systems: An Organizational Perspective. Reading, MA: Addison-Wesley Publishing, 1978.]

    PROJECTOR In 1970 C. L. Meador and D. N. Ness developed PROJECTOR to support financial planning. The system included forecasting and optimization models. It was used in 1974 by a New England manufacturing company to investigate the acquisition of a new subsidiary. [Keen, P.G.W. and M. S. Scott-Morton. Decision Support Systems: An Organizational Perspective. Reading, MA: Addison-Wesley Publishing, 1978.]

    PROSPECTOR acted as a consultant to aid exploration geologists in their search for ore deposits. Given field data about a geological region, it estimated the likelihood of finding particular types of mineral deposits. The system could assess the potential for finding a variety of deposits, including massive sulfide, carbonate lead/zinc, porphyry copper, nickel sulfide, sandstone uranium, and porphyry molybdenum deposits. Expertise was based on 1) geological rules which form models of ore deposits, and 2) a taxonomy of rocks and minerals. PROSPECTOR used a combination rule-based and semantic net formalism to encode its knowledge and based its inferences on the use of certainty factors and the progagation of probabilities associated with the data. This production prototype was implemented in INTERLISP and developed by SRI International. [1. Gaschnig, J. PROSPECTOR: an expert system for mineral exploration. In D. Michie (ed.)Introductory Readings in Expert Systems, Gordon and Breach, Science Publishers, 1982., 2. Duda, R. O. and Reboh, R. AI and decision making: the PROSPECTOR experience. In W. Reitman (ed.) Artificial Intelligence Applications for Business, Norwood, N.J.: Ablex, 1984.]

    RealPlan is an off-the-shelf Real Estate Investment DSS. It performs many of the typical operations required for making real estate acquisition, improvement, and divestment decisions. These include detailed income, expense, and cash flow projections. The portfolio manager creates a table of potential applications for each property. RealPlan then uses a search algorithm to evaluate the most advantageous selection and the timing of actions in terms of cash flows. [Trippi, R.R. A Decision Support System for Real Estate Investment Portfolio Management. Information and Management, Vol. 15, No. 6, December 1988.]

    TAX ADVISOR assists an attorney with tax and estate planning for clients with large estates (greater than $175,000). The system collects client data and infers actions the clients need to take to settle their financial profile, including insurance purchases, retirement actions, transfer of wealth, and modifications to gift and will provisions. TAXADVISOR uses knowledge about estate planning based on attorneys' experiences and strategies as well as more generally accepted knowledge from textbooks. The system uses a rule-based knowledge representation scheme controlled by backward chaining. TAXADVISOR is implemented in EMYCIN. It was developed at the University of Illinois, Champaign-Urbana, as a PhD dissertation and reached the stage of a research prototype. [1. Michaelsen, R.A knowledge-based system for individual income and transfer tax planning, PhD thesis, University of Illinois, Accounting Dept,., Champaign-Urbana, 1982., 2. Michaelsen, R. and Michie, D. Expert systems in business. Datamation, pp. 240-246, November 1983., 3. Michaelsen, R. An expert system for federal tax planning. Report, University of Nebraska, 1984.]

    XCON configures VAX 11/780 computer systems. From a customer's order it decides what components must be added to produce a complete operational system and determines the spatial relationships among all of the components. XCON outputs a set of diagrams indicating these spatial relationships to technicians who then assemble the VAX system. XCON handles the configuration task by applying knowledge of the constraints on component relationships to standard procedures for configuring computers. The system is noninteractive, is rule-based, and uses a forward chaining control scheme. XCON is implemented in OPS5 and was developed through a collaboration between researchers at Carnegie-Mellon University and Digital Equipment Corporation (DEC) in Hudson, Massachusetts. This commmercial expert system configures VAX computers on a daily basis for DEC and is the largest and most mature rule-based expert system in operation. (eXpert CONfigurer of VAX 11/780 computer systems) [1. 1. McDermott, J. R1's formative years. AI Magazine, vol.2, no.2, 1981. 2. Bachant, J. and McDermott, J. R1 revisited: four years in the trenches. AI Magazine, vol.5, no.3, Fall 1984. 3. McDermmot, J. Building expert systems. In W. Reitman (ed.) Artificial Intelligence Applications for Business, Norwood, N.J.: Ablex, 1984.