A Spatio-temporal Taxonomy for the Representation of Spatial Set Behaviours Marius Thériault 1 , Christophe Claramunt 2 and Paul Y. Villeneuve 1 1 Laval University, Planning and Development Research Centre, FAS 16 th floor, Quebec, G1K 7P4, Canada {Marius.Theriault@ggr.ulaval.ca, Paul.Villeneuve@crad.ulaval.ca} 2 The Nottingham Trent University, Department of Computing, Burton Street, Nottingham, NG1 4BU, UK clac@doc.ntu.ac.uk Abstract. Currently, most models proposed for spatio-temporal databases describe changes that involve independent entities. However, many dynamic applications need new models to relate evolution of spatial entities linked by common properties and constraints or relationships. In transportation GIS, an activity-event matrix describes individual entity behaviours, travel activities and routes on a transportation network. On the other hand, modelling disaggregate travel choices behaviour for several entities implies the identification of new mechanisms to describe the evolution of their joint spatial distribution. This paper introduces and describes the concept of sets of geographical entities needed for the analysis of travel behaviour in metropolitan areas. We propose a taxonomy for the description of the evolution of entity sets in space and the selection of appropriate statistical indexes to analyse their geographical patterns. Such a framework may become a reference for the development of spatio-temporal database representations of spatial patterns evolution. 1 Introduction The philosophical revolution of the 18th century introduced a new paradigm of the universe where space and time are perceived as relative components allowing experimentation and perception of the environment. That complements the classical absolute space and time concepts of reality. Locating transformations into their spatio-temporal context is often an essential condition for the full understanding of the real world evolution. Spatio-temporal measurements complement the description of facts and changes, as well as they are used to formulate hypotheses about events and processes, a mandatory requirement to explain evolution using scientific models.