International audience ABSTRACTGeographic knowledge has always proved its value as an instrument underlying reasoning in order to define effective solutions for the management of territories . In many domains, such as urban and environmental planning, it is essential to start processes of analysis and reasoning based on geographic knowledge and then on geolocation, in its broadest sense. In this context, in addition to facts and spatial relationships, knowledge portfolios include geolocalized rules. Statements like "In England,one drives left", "In Canada the majority of the population lives along the border with the United States" or even "The more to North, the colder," recalls that specific rules contain an explicit reference to the space which becomes a fundamental parameter.The nature of these rules is very wide and there are several categories through which one can classify them: (i) the rules of physical geography (e.g. climatogy, floods, volcanic eruptions, diffusion mechanisms, hydrology, etc.), (ii) the rules of human and economic geography (demography, sociology, economics, transport) and (iii) the legal rules (constitutions, laws on urban and environmental planning, the, etc).These rules, sometimes metarules,can come from experts, from spatial data mining algorithms, from documents, from crowdsourcing. However due to their heterogeneity, modeling them through natural languages is a complex task. In fact, the traditional IF-THEN form does not seem totally adapted to shape and manage space geometry and topology. The aim of this work is therefore to introduce some specifications regardingthis set of rule s and propose a model to encode geolocated rules. On the whole, bearing in mind that there may be exceptions, exemptions and offenders, i.e., places where the validity of these rules is no longer guaranteed.