Abstract

The annotation of georeferenced information objects is related to the annotation of the content of the containing spatial regions. Not all spatial features contained in the regions, however, present characteristic attributes of the region described. In this paper, we present a method designed to select prominent spatial features in a region in order to improve the annotation of the region. The method is demonstrated on an artificial dataset and preliminary results show that the method results in a reduction of the number of terms typically describing a region which is statistically significant.