Metagraphs: A tool for modeling decision support systems

Most decision support systems (DSS) contain stored data, data analysis procedures, and decision models. However, many DSS have grown to the point that the average end user is presented with a bewildering array of information resources that are difficult to manage in an effective manner. As a result users often gravitate to a few familiar models and are unaware of the data resources available to them and how these resources relate to the various models. For example, they may think that a model requires data that is unavailable, when in fact that data has recently been added to the data base or could be calculated from another model. Or they may believe that all of the data needed to execute a set of models is available and find out well into the analysis that it is not. Existing tools for DSS design do not provide an effective and comprehensive foundation for modeling all the components of a DSS, or for addressing all the important DSS analysis and design issues. In this paper the authors show how a new graph-theoretic structure, called a metagraph, can be used as a unifying basis for addressing many important questions in DSS development and use.