Characterization of urban risks in the press applying text mining for the enrichment of open data

News is freely spread and widely available to Internet users much more easily than traditional media. In the news, we can find an infinite number of hidden “minor data,” that can provide valuable information not collected in other sources of information. In this context, we have been interested in analyzing and characterizing the urban risks contained in the Uruguayan open newspapers using text mining techniques. This proposal makes it possible to create a news corpus based on risk events included in open data. The corpus covers 2003-2019 and is built from the digital open newspapers El Eco Digital, Montevideo Portal, and La Red 21. Various text mining techniques are applied to this corpus using the QDA-MinerLite software and the Python language (concretely, through the Scattertext library) to identify, characterize, and discover insights on these events. The corpus processing results help enrich the existing open data on risks in Uruguay, incorporating information on their effects, actors, and associated interventions.