*********************************************************** DSS News D. J. Power, Editor December 7, 2003 -- Vol. 4, No. 25 A Bi-Weekly Publication of DSSResources.COM ************************************************************ Post to the Free Bulletin Board at DSSResources.COM ************************************************************ Featured: * DSS Wisdom from "2001: A Space Odyssey" * Ask Dan! - What are major research needs and questions related to building and using model-driven DSS? * What's New at DSSResources.COM? * DSS News Releases ************************************************************ This year, 2003, is the 35th anniversary of the film "2001: A Space Odyssey" by Stanley Kubrick and Arthur C. Clarke. The paranoid, artificially intelligent HAL reminds us of one reason why the Decision Support Systems (DSS) vision keeps a person as the "dominant actor" in an important decision-making process. To quote HAL, "Look, Dave, I can see you're really upset about this. I honestly think you ought to sit down calmly, take a stress pill, and think things over. I know I've made some very poor decisions recently, but I can give you my complete assurance that my work will be back to normal. I've still got the greatest enthusiasm and confidence in the mission and I want to help you." -- HAL (Heuristic and Algorithmic Computer) 9000 For more information check http://www.filmsite.org/ ************************************************************ Put your ad here! Help support DSS News ************************************************************ Ask Dan! What are major research needs and questions related to building and using model-driven DSS? by Dan Power and Ramesh Sharda This Ask Dan! column is a step along a path toward a more rigorous academic journal article tentatively titled "Studying Model-Driven DSS: Research Needs and Directions". Dan Power (Univ. of Northern Iowa and DSSResources.COM) and Ramesh Sharda (Oklahoma State Univ.) are collaborating on this ambitious review and analysis. On February 3, 2002, Dan Power's Ask Dan! column broadly explored "What are the 'hot' DSS research topics?" This column is an expanded discussion focused on model-driven DSS. Even after fifty years of research by economists, psychologists, operations researchers and management scientists we have only just begun to understand the behavioral and technical challenges of designing, developing and implementing model-driven DSS. Given the growing complexity and uncertainty in many decision situations, helping managers use quantitative models to support their decision-making and planning should be a "hot" topic. By definition one or more quantitative models are the dominant component that provides the primary functionality of a model-driven decision support system. Also, by definition a model-driven DSS is designed so a user can manipulate model parameters to examine the sensitivity of outputs or to conduct a more ad hoc "what if?" analysis. Two characteristics differentiate a model-driven DSS from a decision analytic or operations research special decision study: 1) the model(s) in a model-driven DSS are made accessible to a non-technical specialist like a manager through an easy to use interface and 2) the DSS is intended for some repeated use in the same or a similar decision situation. The general types of models used in model-driven DSS include algebraic and differential equation models, analytical hierarchy, decision matrix and tree, multi-attribute and multi-criteria models, forecasting models, network and optimization models, Monte Carlo and discrete event simulation models, and behavioral models for multi-agent simulations (cf., Power, 2002). A number of previous Ask Dan! columns have discussed one or more of these types of models, including AHP, behavioral models, and simulation models. Models in a model-driven DSS should provide a simplified representation of a situation that is understandable to a decision maker. Early versions of model-driven DSS were called model-oriented systems by Alter (1980, p. 92) and category E DSS by Bonczek, Holsapple and Whinston (1981, p. 66). Some authors have called them computationally oriented DSS. Behavioral and technical research on model-driven DSS needs to address many unresolved issues associated with construction of the actual quantitative models, storage and retrieval of data needed by different types of models, communication of parameters among models and among DSS components, multi-participant interaction in model use and value elicitation, and the impact of user interface design alternatives on model-driven DSS effectiveness and ease of use. Also, researchers need to investigate issues associated with building, deploying and using Model-driven DSS. This broad listing of needs seems daunting, but it suggests more specific research issues and questions for further research. The spectrum of possible model-driven DSS is broad and as we learn more the implementation of model-driven DSS expands to incorporate new decision situations and/or new modeling approaches. The remainder of this discussion of model-driven DSS research needs is divided into behavioral and technical questions that seem to warrant further study. Behavioral Questions: The following list of research questions focuses on topics related to understanding the behavioral impact of model-driven DSS. The questions seem largely unresolved in the current literature, but we are identifying relevant prior research. B1. Are users of a model-driven DSS who understand the model more likely to appropriately use the results? Are they more frequent users? Do they have more confidence in the results? The presumption has been that managers need to understand the quantitative model to benefit from using a model-driven DSS. Inadequate research has investigated this topic. B2. Do some users of a model-driven DSS attempt to bias or manipulate the parameters they can change to yield specific results? If so, what types of users and when? What conditions impact biased use of a model-driven DSS? The general perception is that some decision makers will bias an analysis. This is supported by empirical research linked to behavioral decision theory. The phenomenon has not been adequately explored in the context of model-driven DSS. B3. Can some design alternatives and value elicitation methods reduce the occurrence of biased decision behavior? Research associated with manual elicitation of subjective probabilities and values suggests that de-biasing can occur and that some techniques produce normatively better results than other. B4. Does a "customized" user interface impact the subsequent use of a model-driven DSS? How much "customization" is needed and possible? What should the user interface software store from its interaction with a regular user? Personalization of web portals and other Web interfaces is generally considered as desirable and some authors have speculated that because of individual differences among DSS users that a customized interface would improve usability, frequency of use and effectiveness of the model-driven DSS. B5. What is the impact of making decisions with a model-driven DSS in a "highly realistic", simulated decision environment on a person's actual decision making in the "real" decision environment? The increased capability of developing graphical, immersive, "highly realistic" decision situations creates new challenges and opportunities. Some research has begun to examine the impact of realistic decision training using decision support on actual decision making, but much more needs to be done. B6. How can Visual Interactive Simulation (VIS) be used more effectively to examine the consequences of alternative decision strategies and policies. Can a "realistic simulation of a specific firm" enhance firm-specific decision-making? VIS has used Monte Carlo simulation and some technical issues remain, but the impact of this type of simulation deserves more investigation. B7. What behaviors can and should be predicted using economic decision markets? A number of innovative model-driven decision support tools have been developed and investigated, including decision markets and scenario databases (cf., Lang 2003). The limits and possibilities of these approaches need more investigation. Recent controversy over using a decision market to forecast events in the Middle East by the U.S. Defense Department indicates this issue needs investigation. Technical Questions: Technical issues related to building and using model-driven DSS have not been resolved. Rapid technology innovation creates new challenges for researchers interested in more technical research questions. The following questions are not adequately addressed in the current literature. T1. What technology advances are needed to develop the next generation of model-driven DSS generators, especially for creating real time, model-driven decision support systems? In this context, "real time" refers to a contemporaneous analysis using a model-driven DSS while data about events is being received and displayed. The technologies need to provide speed in model execution and updating of data used by the model-driven DSS. T2. How can "real-time" model-driven DSS be interfaced with "real-time" data-driven DSS to improve decision making in situations like routing of an emergency vehicle or selecting passengers to "bump" when a plane is canceled? Real-time data collection and storage issues need more investigation, as do technical issues associated with providing model-driven DSS for use in such an environment. T3. Is a specific extensible mark-up language (XML) needed for communicating data about model parameters? If so, what mark-up tags can create a core for communicating data to various models? Some exploratory research has been conducted on creating a decision support mark-up language (DSML). The varying terminology in use and the variety of categories of DSS suggest that it may be advantageous to have more narrowly defined XML for specific types of models like optimization or discrete event simulation. T4. Can the Uniform Modeling Language (UML) help developers and users of model-driven DSS better understand general categories of model-driven DSS like resource allocation, sourcing and estimating? Identifying and modeling processes like resource allocation or scheduling using UML have been explored, but the possibilities seem promising and more should be done to specify decision processes where quantitative models might be useful. Some processes that should be better defined include 1) assignment (of tasks, of staff, of resources), 2) capacity planning (also queuing and congestion), 3) estimation of costs, quantities, revenues, 4) evaluation and selection (includes using cost-benefit analysis, financial analysis, multicriteria analysis), 5) location analysis (site selection), 6) routing (vehicles, network packets, people), 7) resource allocation, and 8) sequencing and scheduling (cf., Power, 2002). T5. What is an appropriate analytical framework for aggregating results from multiple models? Can web services provide a technical platform for implementing model aggregation? Aggregating results depends upon why aggregation is desired and upon what model results seem to warrant aggregation. Web services are reusable application components that dynamically interact with each other using standard protocols over the Internet. It has been suggested that web services provide a means of dynamically aggregating model results when that is appropriate. T6. What technical capabilities are most appropriate for developing a collaborative, model-driven DSS? Collaborative building of model-driven DSS and collaborative use of model-driven DSS are both interesting areas for further research. The "how" of supporting collaboration may be the same or it may differ in these two situations. T7. What are the tradeoffs among various model-driven DSS delivery mechanisms such as a web browser, spreadsheet, immersive graphics, or peer-to-peer deployment? What innovative user interfaces should be incorporated in next generation model-driven DSS? When is a 3-dimension mouse or electronic ink most useful? The user interface has always been of enormous importance in building any type of DSS and new technologies may be especially useful in enhancing model-driven DSS interaction with a user. T8. How can developers structure "communities of software agents" that imitate social structures like markets, organizations, or nations to assist decision makers for forecasting and planning? Can multi-agent simulations assist managers in understanding emergent behavior in a particular domain and help predict social trends? Realistic simulations using multiple agents for decision support are an exciting frontier that provides many issues for technical research. T9. What reusable model objects should be developed for use in an object-oriented, model-driven DSS development environment? The range of potential quantitative model components is very large and diverse. We need to investigate what objects will be useful and how they can best be implemented. Conclusions The behavioral issues associated with building and using model-driven DSS have often been avoided by relying on specialists and intermediaries to use complex models for analyses, that approach is limiting. Model-driven DSS developers have much more to learn about the management of models and there is a need for new development environments to advance the state of the art. Model management in distributed computing environments is now a requirement and not just a possibility. Model-driven DSS still need to be distributed more widely in organizations and they need to be used by managers and staff for planning and analysis. The representations used in model-driven DSS for planning also need to be used in real-time data-driven decision support. This Ask Dan! presents an ambitious set of research opportunities for the decision support research community and especially those academic researchers and practitioners interested in using models, optimization and simulation to build model-driven DSS. References Alter, S.L. Decision Support Systems: Current Practice and Continuing Challenge. Reading, MA: Addison-Wesley, 1980. Bonczek, R. H., C.W. Holsapple, and A.B. Whinston. Foundations of Decision Support Systems. New York: Academic Press, 1981. Lang, C., "Professional and Collaborative Decision-Support: Many Ways to Improve Decision-Making", DSSResources.COM, 07/11/2003. Power, D., Decision Support Systems: Concepts and Resources for Managers, Westport, CT: Greenwood/Quorum Books, 2002. Please note: The ideas in this Ask Dan! will be presented and discussed at the pre-ICIS SIG DSS Workshop "Research Directions for Decision Support" Dec. 14, 2003 in Seattle, WA USA. Check http://icis2003.cbe.wsu.edu/ ************************************************************ Tell your friends! Get DSS NEWS free -- send a blank email to dssresources-subscribe@topica.com. ************************************************************ What's New at DSSResources.COM? 12/05/2003 Posted Jessani, R., "Creating an Effective Data-Driven Decision Support System". Check the articles page. 11/28/2003 Posted case by Myron Messak, "Decision Support for Mayfield, NY Fire and Emergency Medical Services", 2003. Check the case studies page. ************************************************************ DSS News Releases - November 24 to December 4, 2003 Read them at DSSResources.COM and search the DSS News Archive 12/04/2003 Competitive intelligence: trends and activities; SCIP annual conference March 22 - 24, 2004 in Boston, MA. 12/03/2003 Sun wins worldwide customers and partners for Java(TM) Enterprise System, announces new pricing for independent software vendors, OEMs, and very small companies. 12/03/2003 Zaplet announces partnership with Northrop Grumman's TASC unit to deliver collaborative BPM solutions to Intelligence and Military Agencies. 12/02/2003 BPM Partners to offer a preview of new industry-specific dashboards at DCI Performance Management Conference. 12/02/2003 Groove Networks webcast addresses distributed project management challenges. 12/01/2003 MicroStrategy provides free BI software to the Teradata University Network. 12/01/2003 Corporate Risk managers have a new tool to assess and manage catastrophe risk with the release of AIR's CATStation. 12/01/2003 BizTools exhibits business intelligence application at Intuit's 2003 IDN conference; Intuit's 2003 IDN conference attracts developers from around the world. 12/01/2003 AutoTrader.com selects MicroStrategy Report Services for enterprise reporting and analysis. 12/01/2003 Countrywide Home Loans looks to MindBox for loan decision support; expands MindBox's ARTEnterprise license to enable automated mortgage underwriting system across entire enterprise. 12/01/2003 Pentagon selects Sterling Management Solutions Corp.'s Performance Measurement/BI and Business Continuity Virtual Command Center (VCC) Dashboard solutions. 11/29/2003 Century 21 Co-founder launches expert systems; new IdeaFisher Expert Systems™ products ask thousands of probing questions, provides insightful answers. 11/28/2003 ABN AMRO insurance arm selects Fair Isaac decision tools for more efficient auto insurance underwriting. 11/26/2003 U.S. Department of Homeland Security looks to academia for help fighting terrorism. 11/26/2003 Tokyo University sets bandwidth record at SC2003 with Juniper Networks; Japanese research team exceeds 7.5 Gbps between Japan and the USA with T320 platform. 11/25/2003 Lombardi Software's TeamWorks 4 selected as finalist for Transform Magazine's Product of the Year 2003 awards. 11/25/2003 SPSS enables full customer view with new predictive text analytics solution. 11/24/2003 Infommersion wins best of COMDEX Las Vegas 2003 award for software; Xcelsius allows users to create real-time, interactive reports based on Excel spreadsheets. ************************************************************ Subscribe to DSSResources.COM. One month $10, six months $25. Visit http://dssresources.com/subscriber/subscriber.html ************************************************************ DSS News is copyrighted (c) 2003 by D. J. Power. Please send your questions to daniel.power@dssresources.com. |