Aggregating Data for Decision Support 111 Henk G. SOL Information Systems Group, Department of Computer Science, Delft University of Technology, Julianalaan 132, Delft, The Netherlands Numerous strategic and tactical decisions in organisations are based on aggregateddata. A simulation-based inquiry of a Multiple Store Company leads to the conclusion that aggrega- tion of data on local decisions does not give insight for taking global decisions. This implies that one has to question the validityof management information produced throughaggrega- tion in numerou~data processingsystems. It also signifiesthat application of models based on definition and behavioural equations can be dangerous. Simulation-based inquiry systems are, however, capable of providing support even in these cir- cumstances. With a simulation-baseddecision support system for global decision-making we may analyze throughdisaggrega- tion the effects of these decisions at local levels. By using a problem-solvingenvironmentbased on system descriptionand simulation,effectivedecisionsupport systemscan be developed efficiently. Keywords: Aggregation,DSS, Simulation North-Holland Decision Support Systems 1 (1985) 111-121 Introduction There is not yet a generally accepted definition of Decision Support Systems (DSS). The change in description of DSS from 'concept' through 'move- ment' to 'bandwagon' clearly illustrates the grow- ing interest in the managerial as well as in the research field for decision support systems. A useful framework for research on DSS is intro- duced in Sprague [1980]. He discusses the perspec- tive of the end-user, the builder and the toolsmith from which a DSS can be viewed. In accordance with this distinction the concept of a DSS-genera- tor is put forward to bridge the gap between general tools and specific DSS. Sprague distinguishes as the main components of a DSS a data base, a model base, and an intermediate software system which interfaces the DSS with the user. Sprague and Carlson [1982] advocate an approach to systems analysis which is intended to identify requirements in each of the three major capability areas of DSS: The approach is based on a set of four user-oriented entities: Representations, Operations, Memory Aids and Control Mechanisms'. This so-called ROMC ap- proach can be placed in the framework proposed by Bonczek et al. [1980]. They replace the compo- nents mentioned by the concepts of a language system (LS), a knowledge system (KS) and a prob- lem processing system (PPS). The language system is the sum of all linguistic facilities made available to the decision-maker by a DSS. A knowledge system is a DSS's body of knowledge about a problem domain. The problem processing system is the mediating mechanism between expressions of knowledge in the knowledge system and expres- sions of problems in the language system. In Sol [1982] we argued that much more atten- tion has to be paid to the process of solving ill-structured problems. ,Problem-solving is an iter- ative modeling process, in which we identify the activities of conceptualization, problem specifica- tion, solution finding and implementation. We make a distinction between: (a) Conceptual and empirical models; (b) Descriptive and prescriptive models. We call a descriptive empirical model an 'under- 0167-9236/85/$3.30 © 1985, ElsevierSciencePubfishersB.V. (North-Holland)