Partially Ordered Preferences Applied to the Site Location Problem in Urban Planning

This paper presents an application that aims at identifying optimal locations based on partially ordered constraints. It combines a tool developed in this project that allows the management of partially ordered constraints and a Geographical Information System (GIS) allowing spatial data mapping and analysis. Experts in urban planning provides constraints, being in our application a combination of legal constraints and preferences expressed by the property developer. As these constraints can hardly be totally ordered because they are not comparable, constraints are partially ordered. The experiment was performed using about 3800 cadastral parcels and 12 different constraints, each parcel being characterised for each constraint using GIS analysis operators. Data are then processed by the program mpropre (Managing PaRtially Ordered PREferences) that provides in output one or several optimal parcels. Results are finally validated by an expert and using ortho-images of the geographic area of interest.