О географической привязке контента текстовых документов

Extracting geographical names from arbitrary text documents is important in the tasks of processing large arrays of documents and linking their content to a specific geographic region. In the simplest form, the model for extracting geographical names from the text looks like a sequence of actions with the text, while at each stage its task is solved. Among these tasks, there are undoubtedly: text parsing, analyzing text elements, processing synonyms and abbreviations, bringing the text elements to normal form from possible word forms and grammar rules, comparing text elements with the elements of dictionaries of geographical names, adding special tags to the text for unambiguous identification geographical names. The proposed work describes a technology that implements the above tasks on the basis of a freely distributed PostgreSQL DBMS. In this case, the standard configuration is used, all the server part settings are performed within the framework of the documented procedures. GeoNames Gazetteer database, Open Street Map (OSM) databases, OKATO and КЛАДР classifications are used as an authoritative database of geographical names.