SpatialML: annotation scheme, resources, and evaluation

摘要: SpatialML is an annotation scheme for marking up references to places in natural language. It covers both named and nominal places, grounding them where possible with geo-coordinates, characterizes relationships among terms of a region calculus. A freely available editor has been developed SpatialML, along several annotated corpora. Inter-annotator agreement on extents 91.3 F-measure corpus SpatialML-annotated ACE documents released by the Linguistic Data Consortium. Disambiguation geo-coordinates 87.93 F-measure. An automatic tagger scores 86.9 F ACE, while disambiguator 93.0 it. Results are also presented two other In adapting extent new domains, merging training data from domain provides best performance.