Evaluating Geographic Information Retrieval

In GIR, as a research field branching from IR, initial evaluation initiatives were unsurprisingly inspired from established IR evaluation models. The GeoCLEF evaluation [6], which I will look into in some detail below, follows such a model. While this model suits the goal of a generic evaluation for all types of GIR systems, other evaluation initiatives forked so that particularly geographic challenges of the task could be more thoroughly focused on. NTCIR [7], with its GeoTime task, looked to include time expressions and had a topic set that resembles more questions that stir up the focus from retrieval, and more on reasoning.