Multi-agent-based system technologies in environmental issues

Abstract

Environmental impact assessment (EIA) is one of the principal indicators for human health evaluation as well as for further issues in occupational medicine, public health and planning. Certain difficulties to evaluate EIA appear due to, on the one hand, the nature of contaminants and health characteristics, and, on the other, the difficulties found in data processing, such as heterogeneity of the information sources, lack of data quality and need to use numerous data mining methods. The application of multi-agent systems (MAS) to EIA helps operating with distributed decentralized information and generating general decisions and solutions. In order to facilitate the workflow of MAS planning and implementation in decision support systems (DSS) applied to EIA, a generalized multi-agent system technology (MAST) is proposed. This workflow is general enough to enable the related MAS to be used in any application area requiring decision making. We demonstrate that by only changing the domain ontology the MAS is easily oriented towards another problem area. The functional organization and the roles executed in the MAS conform to the logical sequence of the data transformation flow, which include information retrieval and fusion, data pre-processing, data mining and modeling, and simulation and decision making. © Springer-Verlag Berlin Heidelberg 2009.