Geoscience Reference
In-Depth Information
- the effects of interactions between city-agents on the spatial and hierarchical
organization of a settlement system (e.g. EuroSim).
However, it is not known how to build a model simultaneously integrating the
three levels involved (those of individuals, cities and systems of cities), and the two
nested stages of emergence associated. The main reason is the semantic change of
the interactions when one passes from one level to the other: the interactions
between territorialized entities relate, as a matter of fact, to another subject than
those between individuals. How to identify spatial entities which make sense in the
segregation field that has emerged at the mesogeographical level, in the case of
Schelling's model, for example? Or in the settlement pattern which has emerged
from the interactions between Pueblo households? And how to formalize
interactions between groups without these groups being identified as autonomous
entities? These difficulties are related to the definition, identification and
delimitation of entities of interest that emerge from the interactions between more
simple entities.
In the field of data processing, these issues of entities' definition are also
fundamental. If we have not mentioned in Chapter 3 the question of the MAUP,
crucial in the dialogue between statisticians and geographers, that is because our
position in this regard coincides with that of Openshaw: “according to one given
domain, if geographers would agree on what constitute objects of research and if
these objects would be defined in a non-arbitrary way, the problem of the spatial
aggregation would disappear” [OPE 81]. When the meaning of the entities is well
specified, the possible differences in results obtained by applying the same method
to two sets of spatial entities delimited differently, rather than being a problem,
reveal the multi-scalar structure of the geographical space [MAT 06]. These issues
of entities' definition are today reactivated with the new data produced by all the
sensors that surround us, RFID chips, RSS feeds, social networks exchanges,
geoweb2, etc. All these data (operational data, declarative data, etc.) constitute a
crude material for which the elementary entities are rarely the entities to be
analyzed: the use of this material needs in a first time to filter the data in order to
reconstruct entities (simple or composite) having a meaning for the question that
was raised. The concern is, for example, to identify stopping places from mobile
phone data and use it to study individual's spatial practices [OLT 12]; to use regions
cited by AFP (Agence France Press) for following a diffusion, etc. The use of an
approach based on an explicit ontology takes its full significance here.
The position taken in this topic has been to browse all the steps associated with a
spatio-temporal problematic, from the critical identification of the concerned entities
to the simulation of the processes underlying the evolution of the phenomenon of
interest. The approaches described in the different chapters most often mobilize
different scientific communities. The point was, and it is the originality of the
