Planned Workshop

Note! This workshop was originally planned to be held in conjunction with AGILE 2017. However, we decided to postpone it to a later date, in order to allow for better preparation and a more timely activation of the community. Please contact the organizers if you like the idea of the workshop and if you are interested in getting involved.


Workshop on Graph Models and Optimization in GI Science

Graphs are mathematical models that are suitable for networks and maps. Therefore, they are of fundamental importance in geographical information (GI) science. This workshop addresses graph models in spatial planning, geographic analysis and cartographic visualization, with the aim of intensifying communication between researchers who work with a similar methodology in these different branches of spatial science. We encourage contributions presenting mathematically rigorous models as well as algorithmic solutions for those problems. Since solving a model usually means optimizing an optimization objective subject to a set of constraints, the workshop primarily addresses optimization algorithms, being they exact, approximate, or heuristic.

On the one hand, optimization approaches are often based on simplified problems with strong constraints, exact solving purpose and basic linear and additive objective functions, which do not perfectly reflect the way geographical issues are tackled by information scientists. On the other hand, GI Sciences show a lack of quantitative and mathematical frames in optimization to model and investigate geographical phenomena. Therefore, we will address the following questions in the workshop: How could it be possible to soften the constraints set fixed in optimization approaches in order to find more accurate and reliable solutions? How can users and decision makers manage to balance computing efficiency and exactness of the solutions?

We will discuss how problems of high computational complexity (e.g., NP-hard problems) in GI science can be tackled. Do we accept suboptimal solutions for the sake of efficiency and, if so, how far away from optimum do we allow them to be? How do we evaluate the results of heuristics? Which role do exact algorithms for NP-hard problems (e.g., mathematical programming solvers) play? 


Jan-Henrik Haunert, University of Bonn

Takeshi Shirabe, KTH, Stockholm, Sweden

Didier Josselin, CNRS, Université d’Avignon et des Pays de Vaucluse, France