In computer science, data modeling has been the core activity of database design for the last twenty years. However, for historical reasons, the GIS community has been slow to adopt this key activity of the information system lifecycle; mostly, GIS specialists have built relational tables with excellent technical knowledge but without formal and rigorous methods like the ones used with other information technologies. Fortunately, in the recent years, it has been changing rapidly with the advent of object-oriented techniques and the increased interest towards systems interoperability. The objective of this paper is to explain how the modeling techniques used in computer science support the brain mechanisms involved in designing a spatio-temporal database and how we can adapt these methods for our specific needs.
We begin with a description of the fundamental concepts underlying the modeling activities with regard to the functioning of the human brain. This description shows to the importance of using both graphical and textual representations during the design of a database. Then, we introduce the design needs which are specific to the spatial and temporal properties of geographic objects. Similarities between spatial and temporal properties are highlighted. Then comes a presentation of two popular modeling techniques (Merise Entity/Relationship and OMT Object-Oriented) with application examples. We go into slightly more details for the third method, that is Modul-R, that we developed specifically for spatio-temporal databases. Then, we make a comparative analysis of these three methods where advantages and disadvantages are given. Finally we conclude with a brief discussion and list future research directions.