A Decision Support System For The Requalification Of Contaminated Sites: A Fuzzy Expert Approach For Comparing Alternative End Uses

An important aspect in the issue of contaminated sites is their use after remediation. The choice of a certain use has got necessarily socio-economic implications that make it more or less suitable for that site. For this reason it is important to insert socio-economic analysis within the studies on contaminated site remediation and to implement approaches that calibrate extent and modality of the remediation on the basis of potential uses of the site after it. In particular it is important to provide decision-makers with tools that offer them the possibility to consider also this aspect, in order to support their decisions.These considerations have led to introduce also a socio-economic module inside the DESYRE, a project that has implemented a software prototype with a Decision Support System for the remediation of contaminated sites. Specific contents and details of the project are reported in a contribution presented in this conference (Samiolo et al.).The specific aim of the socio-economic module in the software is providing the decision makers with a tool that makes possible to compare the different use destinations, outlining possible scenarios linked to alternative uses of the considered site, on the basis of socio-economic considerations (often founded on theories and methods of the spatial analysis) and of local characteristics. Comparing these scenarios it aims to give indications on which use is more suitable and why. The final objective is just to establish which is the “best” use for that site. The term “best” indicates that it is possible to rank the whole different use destinations. This is a typical multicriteria decision making problem and there is not a natural order in a multidimensional space, so it is necessary to find a device to do it. A method to rank them is, fixed an use destination (UD), to choose a method that defines a function from the attribute space in the real line and so obtain a total order induced by the real number one. The scientific literature is reach of several ways to approach this problem. The new researches in Artificial Intelligence propose two different types of methods: the first group is formed by Expert Systems (ES), Fuzzy Expert Systems (FES), Hierarchical methods (AHP), the second one is formed by Data-Mining methods (Neural Networks, Genetic Algorithms, etc.). The first ones are knowledge based systems, that is does not use the past data and permits a real contact with the experts that may allow into the study all the experience matured in years of work in that field. The second ones are called “knowledge-discovery in data base” and are based only on data. By them the methods are able to approximate the function that describes the model. The problem is that they need of a great number of data to be efficient, but in this case we have only data relative to one case, those of two sites located in Porto Marghera (Venice – Italy). When the data are insufficient we have to use the first type. We have decide to use FES for hits efficiency and capacity to approximate the complicate function that describes the problem.