Geographical or spatial information is now included in most of exchanged data. Sometimes, it is directly provided through metadata, but it is very often hidden and it becomes crucial to automatically discover it. Natural Language Processing (NLP) and Data Mining communities have thus merged their efforts in order to extract geospatial information from textual documents, web pages, field data, and so forth. In this way, recent researches take into account the content of documents (e.g. terms) to identify geospatial data or to predict its geographic location.