Unnamed Locations, Underspecified Regions, and Other Linguistic Phenomena in Geographic Annotation of Water-based Locations

摘要

This short paper investigates how locations in or close to water masses in topics and documents (e.g. rivers, seas, oceans) are referred to. For this study, 13 topics from the GeoCLEF topics 2005-2008 aiming at documents on rivers, oceans, or sea names were selected and the corresponding relevant documents retrieved and manually annotated.rnResults of the geographic annotation indicate that i) topics aiming at locations close to water contain a wide variety of spatial relations (indicated by different prepositions), ii) unnamed locations can be generated on-the-fly by referring to movable objects (e.g. ships, planes) travelling along a path, iii) underspecified regions are referenced by proximity or distance or directional relations. In addition, several generic expressions (e.g. "in international waters') are frequently used, but refer to different underspecified regions.