The huge diffusion of social networks has made available an unprecedented amount of publicly-available user-generated data,which may be analyzed in order to determine people’s opinions and emotions. In this paper we investigate the use of Bidirectional Encoder Representations from Transformers(BERT) models for both sentiment analysis and emotion recognition of Twitter data.We define two separate classifiers for the two tasks and we evaluate the performance of the obtained models on real-world tweet datasets. Experiments show that the models achieve an accuracy of 0.92 and 0.90 on, respectively, sentiment analysis and emotion recognition.

Emotion and sentiment analysis of tweets using BERT

Andrea Chiorrini;Claudia Diamantini;Alex Mircoli;Domenico Potena
2021

Abstract

The huge diffusion of social networks has made available an unprecedented amount of publicly-available user-generated data,which may be analyzed in order to determine people’s opinions and emotions. In this paper we investigate the use of Bidirectional Encoder Representations from Transformers(BERT) models for both sentiment analysis and emotion recognition of Twitter data.We define two separate classifiers for the two tasks and we evaluate the performance of the obtained models on real-world tweet datasets. Experiments show that the models achieve an accuracy of 0.92 and 0.90 on, respectively, sentiment analysis and emotion recognition.
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: http://hdl.handle.net/11566/289290
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 11
  • ???jsp.display-item.citation.isi??? ND
social impact