Abstract (EN)

A growing concern on the quality of urban environments has recently developed, due to the effects of pollution on human health and on socio-economic systems and particularly caused by urban production mechanisms. Emission-reducing policies and new analytical models have been increasingly used to cope with such new environmental threats. Therefore, there is the need of integratine traditional quantitative (statistical, data managing, geographical, etc.) techniques .with more complex tools, so as to provide decision makers with greater support to elaborate effective socio-environmental strategies. ECOURB research project, in which this paper is located, just deals with the identification of models to support urban environmental decisions. It is developed coherently with a general decision-support ICT-oriented architecture, through the analysis of multi-agent processes and the identification of variables, parameters, indicators useful in such system architectures. The present paper shows some results of this research project, in the building up of hybrid scenarios for the management of urban microclimates. Particularly, the reference context is represented by the management of indoor spaces in inhabited urban areas. Reflections are carried out on the case-study of public condo housing owned by the Institute of popular housing (IACP) in Bari, starting from a knowledge base collected through questionnaires. Such reflections take into particular account the role, the actions and the contributions of agent distributed in the inhabited context, in mutual, social and external relationships toward the reference physical environment (heat/cold management, urban heat islands, air quality etc.). The multi-agent approach allows the consideration of the complexity of such interactive relationships, so as to use them for the building up of social and environmental management models. Several agent types are taken into consideration, such as routinary human and artificial agents, as well as intelligent agents, so following taxonomical and logical approaches deriving from computer science and multi-agent theory. The modelling approach is characterized in the identification of single agents involved and in the definition of some logic rules of relation and behaviour. However, such model represents a preliminary study outcome, to be developed in the carrying out of the project with the extension and the deepening of the research project.