A Hybrid Approach to Information and Decision Support Systems: Hazardous Substances and Industrial Risk Management

This paper describes a large-scale, model-based decision support system with embedded Artificial Intelligence (AI) technology, and a largely symbolic color graphics user interface. Designed for industrial risk assessment and the management of hazardous substances, its primary purpose is to allow the efficient use of methods of systems analysis and information management and to provide a powerful tool in the hands of planners, managers, policy and decision makers. This new generation of model-based decision support system should lead to a more informed, structured, comprehensive and interdisciplinary management of hazardous substances. To facilitate access for the non-technical user, and for more experimental and explorative use, it proved necessary to build much of the accumulated knowledge on the subject areas into the user interface. Thus, the interface incorporates elements of knowledge-based expert systems that are capable of assisting any non-expert user to select, set up, run, and interpret complex technical software. By providing a coherent integrated user interface, the interactions between different models, their data bases, and software for display and analysis become transparent and the substantive information basis for decision support can be considerably enlarged. As a central component of this hybrid decision support tool, a heterarchical, object-oriented information system on hazardous substances has been incorporated. It provides aggregated as well as detailed information for non-specialist users and allows easy modular updating with automatic consistency checking. To cover all the different viewpoints under which one can group hazardous substances (e.g., chemical compounds, toxicity, etc.) an object-oriented heterarchy of viewpoints, groups, and subgroups, down to the elementary substances has been set up. Each object is activated when accessed, provides its descriptive data, refers to and derives data from other objects. In addition, every object statically inherits all the descriptive data and activities from all its predecessors in the heterarchy