Daniel R. Dolk

Knowledge Representation for Model Management Systems

This paper examines the concept of a model management system, what its functions are, and how they are to be achieved in a decision support context. The central issue is model representation which involves knowledge representation and knowledge management within a database environment. The model abstraction structure is introduced as a vehicle for model representation which supports both heuristic and deterministic inferencing as well as the conceptual/external schema notion familiar to database management. The model abstraction is seen as a special instance of the frame construct in artificial intelligence. Model management systems are characterized as frame-systems and a database implementation of this approach is described.

Model integration and a theory of models

Model integration extends the scope of model management to include the dimension of manipulation as well. This invariably leads to comparisons with database theory. Model integration is viewed from four perspectives: Organizational, definitional, procedural, and implementational. Strategic modeling is discussed as the organizational motivation for model integration. Schema and process integration are examined as the logical and manipulation counterparts of model integration corresponding to data definition and manipulation, respectively. A model manipulation language based on structured modeling and communicating structured models is suggested which incorporates schema and process integration. The use of object-oriented concepts for designing and implementing integrated modeling environments is discussed. Model integration is projected as the springboard for building a theory of models equivalent in power to relational theory in the database community.

Model management and structured modeling: the role of an information resource dictionary system

Models have historically occupied an ambiguous position within organizations. Management acceptance of management science and operations research models for decision-making has lagged far behind technical advances in these areas. Structured modeling has emerged as a unifying approach to the modeling process with potential to reduce this ambiguity. Structured modeling is primarily oriented to the individual, however. A way of incorporating structured modeling into the organizational framework via existing information resource channels is discussed. A relational model is presented for an information resource dictionary system (IRDS). This IRDS model is then extended to accommodate representation of structured models. This extension of the IRDS can answer queries about structured modeling as well as model instances. The concept of an active IRDS is introduced and an example presented of how an active IRDS can be linked with an optimization algorithm. The conclusion is that the IRDS is a suitable vehicle for incorporating model management and structured modeling as part of an organizations information resource management environment.

A generalized model management system for mathematical programming

This paper examines mathematical programming software in the context of model management and decision support. The concept of a model management system (MMS) is introduced and compared to traditional modeling systems. An MMS is seen as a much more generalized software system that requires the confluence of existing operations research, database management, and artificial intelligence techniques. By incorporating powerful, abstraction-based representation structures, an MMS can support multiple levels of model abstraction, only one of which corresponds to traditional, solution-oriented modeling software. The database structures required to implement a knowledge-based MMS are discussed and a prototype system for mathematical programming, the Generalized eXperimental Math Programming system (GXMP), is described. An algebraic language developed for use in GXMP is described in detail.

Design principles for virtual worlds

In this research note, we examine the design, development, validation, and use of virtual worlds. Our purpose in doing so is to extend the design science paradigm by developing a set of design principles applicable to the context of virtual environments, particularly those using agent-based simulation as their underlying technology. Our central argument is that virtual worlds comprise a new class of information system, one that combines the structural aspects of traditional modeling and simulation systems in concert with emergent user dynamics of systems supporting emergent knowledge processes. Our approach involves two components. First, we review the characteristics of agent-based virtual worlds (ABVWs) to discern design requirements that may challenge current design theory. From this review, we derive a set of design principles based on deep versus emergent structures where deep structures reflect conventional modeling and simulation system architectures and emergent structures capture the unpredictable user-system dynamics inherent in emergent knowledge processes, which increasingly characterize virtual worlds. We illustrate how these design challenges are addressed with an exemplar of a complex mirror world, a large-scale ABVW we developed called Sentient World. Our contribution is the insight of partitioning ABVW architectures into deep and emergent structures that mirror modeling systems and emergent knowledge processes respectively, while developing extended design principles to facilitate their integration. We conclude with a discussion of the implications of our design principles for informing and guiding future research and practice.

Model Integration and Modeling Languages: A Process Perspective

Development of large-scale models often involves-or, certainly could benefit from-linking existing models. This process is termed model integration and involves two related aspects: 1 the coupling of model representations, and 2 the coupling of the processes for evaluating, or executing, instances of these representations. Given this distinction, we overview model integration capabilities in existing executable modeling languages, discuss current theoretical approaches to model integration, and identify the limiting assumptions implicitly made in both cases. In particular, current approaches assume away issues of dynamic variable correspondence and synchronization in composite model execution. We then propose a process-oriented conceptualization and associated constructs that overcome these limiting assumptions. The constructs allow model components to be used as building blocks for more elaborate composite models in ways unforeseen when the components were originally developed. While we do not prove the sufficiency of the constructs over the set of all model types and integration configurations, we present several examples of model integration from various domains to demonstrate the utility of the approach.

Model management in organizations

Abstract The premise that the personal computer/spreadsheet explosion will result in the evolution of model management within organizations is explored. The authors use Nolans stage model of organizational data processing activity as a basis for discussing the nature of change in organizations as local computing capability proliferates. The mainframe era resulted in the recognition of data as a resource and gave rise to data administration. The authors expect that the personal computer era and the accompanying spreadsheet explosion will lead to the recognition of models as a valuable and manageable resource. The role of model administration within organizations is discussed as are software tools for supporting this functional activity. The information resource encyclopedia, an extension of the traditional data dictionary concept, and the model management system are introduced as integral components for supporting model administration. An example is presented to suggest an integrative approach for implementing an MMS in a spreadsheet environment.

Integrated model management in the data warehouse era

Abstract Integrated modeling environments have historically been viewed as extensions of database management systems. Now that the WorldWideWeb has commandeered the computing landscape, a distributed perspective is required for model management. Our approach to distributed integrated modeling combines a data warehouse of decision metrics, a model warehouse of decision model structural representations, a conversion process to represent models in object-oriented form using the unified modeling language (UML), and component-based software plug-ins which augment typical cross-tabulations analysis with more sophisticated solvers from optimization, statistics, and other modeling paradigms.

A relational information resource dictionary system

A relational implementation of IRDS using SQL demonstrates how the flexibility of the relational environment enhances the extensibility of the IRDS while at the same time providing more powerful dictionary capabilities than are typically found in relational systems.

An introduction to model integration and integrated modeling environments

Abstract Integrated modeling systems provide support for the definition, manipulation, and control of mathematical models throughout the entire modeling life cycle. Model integration is a particularly crucial operation which requires thinking about “modeling in the large”, and which extends the scope of model management research to include manipulation as well as definition. Several aspects of model integration are identified and briefly described with respect to the problems they raise for constructing integrated modeling environments. Relevant work in these areas is cited. A brief introduction to each of the papers in this special issue is provided within the context established.