Mining Social Networks to Detect Traffic Incidents

VALLEJOS, SEBASTIAN; ALONSO, DIEGO G.; CAIMMI, BRIAN; BERDUN, LUIS; ARMENTANO, MARCELO GABRIEL; SORIA, ÁLVARO

INFORMATION SYSTEMS FRONTIERS

SPRINGER

Año: 2020 p. 115 - 134

1387-3326

Social networks are usually used by citizens to report or complain about traffic incidents that affect their daily mobility. Automatically finding traffic-related reports and extracting useful information from them is not a trivial task, due to the informal language used in social networks, to the lack of geographic metadata, and to the large amount of non traffic-related publications. In this article, we address this problem by combining Machine Learning and Natural Language Processing techniques. Our approach (a) filters publications that report traffic incidents in social networks, (b) extracts geographic information from the textual content of the publications, and (c) provides a broadcasting service that clusters all the reports of the same incident. We compared the performance of our approach with state of the art approaches and with a popular traffic-specific social network, obtaining promising results.