Generating geographical location descriptions with spatial templates: a salient toponym driven approach

Natural language descriptions of geographical locations are used frequently in daily life and there is a motivation to create systems that generate such automatically, for purposes as documentation where events have taken place, person located, photos were plants animals located. Typically location combine references named features with vague spatial relational terms, near, north at relate the features. Here we describe system generating descriptions, combines templates, model applicability different relations relative reference location, toponyms vicinity described selected according aspects salience. The retrieved from gazetteer service based on OpenStreetMap which hierarchical feature classification scheme facilitate selection distinctiveness their types other advantages approach demonstrated user study, an existing state art baseline approaches include manually created captions automated methods two widely photo captioning systems.