Automatic Acquisition of Fuzzy Footprints

Gazetteer services are an important component in a wide variety of systems, including geographic search engines and question answering systems. Unfortunately, the footprints provided by gazetteers are often limited to a bounding box or even a centroid. Moreover, for a lot of non–political regions, detailed footprints are nonexistent since these regions tend to have gradual, rather than crisp, boundaries. In this paper we propose an automatic method to approximate the footprints of crisp, as well as imprecise, regions using statements on the web as a starting point. Due to the vague nature of some of these statements, the resulting footprints are represented as fuzzy sets.