Processing and analysis of messages for geocoding of anomalous cases

The purpose of the research was to study and implement methods for collecting and processing data obtained from social networks to extract useful information about emergency situations. Information extracted from social media posts can help detect the fact of an emergency and establish its approximate geographical location based on the text of the record reporting the incident. Tools were designed and developed that implement the collection and analysis of records from the VK social network communities. A tool has been created to extract the geographical location described in the analyzed text. Basic natural language processing methods, named entity recognition methods, frequency analysis and geocoding tools are used to extract information. The developed programs may have practical significance for public authorities, security services and other interested organizations in the field of ensuring the safety and protection of the population.

Authors: А. M. Lopushanskiy, Ya. A. Bekeneva

Direction: Informatics, Computer Technologies And Control

Keywords: social networks, message processing, location, geocoding, NER (named entity recognition)

View full article