In this paper we present our contribution to the TREC 2019 Incident Streams track. We submitted four runs to the 2019-B edition of this task. Our main goal is to evaluate the effectiveness of sentiment analysis and information retrieval techniques to automatically detect and prioritize incidents on social media streams. Here, we describe these techniques and show the results obtained.
Revista: TREC 2019
Autores: Javi Fernández, Fernando Llopis, Patricio Martínez-Barco, José M. Gómez
URL: https://trec.nist.gov/pubs/trec28/papers/UAGPLSI.IS.pdf