Descripción
|
|
---|---|
This article describes the system developed by the Grupode Tecnolog ?a del Habla at Universidad Polit ?ecnica de Madrid, Spain(GTH-UPM) for the competition on sentiment analysis in tweets: TASS2019. The developed system consisted of three classifiers: a) a systembased on feature vectors extracted from the tweets, b) a neural-basedclassifier using FastText, and c) a deep neural network classifier usingcontextual vector embeddings created using BERT. Finally, the averagedprobabilities of the three classifiers were calculated to get the final score.The final system obtained an averaged F1 of 48.0% and 48.4% for thedev set on the mono and cross tasks respectively, 46.0% and 45.0% forthe mono and cross tasks for the test set | |
Internacional
|
Si |
Nombre congreso
|
Iberian Languages Evaluation Forum @ SEPLN |
Tipo de participación
|
960 |
Lugar del congreso
|
Bilbao |
Revisores
|
Si |
ISBN o ISSN
|
1613-0073 |
DOI
|
|
Fecha inicio congreso
|
02/09/2019 |
Fecha fin congreso
|
03/09/2019 |
Desde la página
|
579 |
Hasta la página
|
588 |
Título de las actas
|
Proceedings of the Iberian Languages Evaluation Forum @SEPLN |