Home // International Journal On Advances in Software, volume 13, numbers 3 and 4, 2020 // View article

Using Natural Language Processing for Extracting GeoSpatial Urban Issues Complaints from TV News

Authors:
Rich Elton Carvalho Ramalho
Anderson Almeida Firmino
Cláudio De Souza Baptista
Ana Gabrielle Ramos Falcão
Maxwell Guimarães de Oliveira
Fabio Gomes de Andrade

Keywords: Geosocial network; NLP; Urban Issues; Crowdsourcing.

Abstract:
Citizens as sensors enable society to discuss urban issues. Although some geosocial networks have been developed in recent years to enable citizens to report many types of urban problems, the engagement of the users of these networks usually decreases in time. Hence, many relevant issues are not posted reducing the effectiveness of these networks. Aiming to overcome this limitation, this article proposes an approach in which urban issues are automatically detected from a television news program. The proposed solution uses geoparsing and Natural Language Processing techniques to geocode and classify the identified complaints. The results are published in the Crowd4City geosocial network that deals specifically with urban issues. Finally, our method was evaluated using data from a real news TV program in Brazil. Our results indicated 59.8% of success in extracting text and location from the video news.

Pages: 229 to 239

Copyright: Copyright (c) to authors, 2020. Used with permission.

Publication date: December 30, 2020

Published in: journal

ISSN: 1942-2628