Abstract

The automatic extraction of geospatial information in real time, from free speech, will an important impact on disruptive technologies, such as AI assistants and search engines. Models capable of decoding discourse and then transforming it into structured data addresses the solution of challenging problems such as: the automatic identification of phrases refering geospatial entities; the handling of synonyms referring to the same location (ambiguity); the taxonomic characterization of locative expressions; as well as the automation of complex processes of natural language interpretation to determine of geographical coordinates, maps and databases. The presented project aims to automate georeferencing processes of digital documents to extract georeferenced knowledge. The methodological proposal includes compilation of a corpus, linguistic characterization and algorithms.