Stance detection on social media has attracted a lot of attention in the last few years, as opinionated posts are an invaluable source of information which can possibly be exploited in dedicated systems. This is especially true in the case of particularly polarizing topics for which there is no clear consensus among population. In this paper, we focus on one of these topics, namely the EU digital COVID certificate (also known as Green Pass), with the objective of uncovering the stance towards it in a specific time period for the Italian Twitter community. To this aim, we first tested some classifiers for determining the most suitable one in terms of performance and complexity for the stance detection problem under consideration. Then, we compared several approaches aimed at counteracting the occurrence of concept drift, i.e., that phenomenon for which the characteristics of the dataset vary over time, possibly resulting in a degradation of classification accuracy. Our experimental analysis suggests that updating the classifier during the stance monitoring campaign is crucial for maintaining a satisfactory level of performance. Finally, we deployed our system to monitor the stance on the topic of Green Pass expressed in tweets published from July to December 2021 and to obtain insights about its

Online Monitoring of Stance from Tweets: The case of Green Pass in Italy

Alessandro Bondielli;Giuseppe Cancello Tortora;Pietro Ducange;Armando Macrì;Francesco Marcelloni;Alessandro Renda
2022

Abstract

Stance detection on social media has attracted a lot of attention in the last few years, as opinionated posts are an invaluable source of information which can possibly be exploited in dedicated systems. This is especially true in the case of particularly polarizing topics for which there is no clear consensus among population. In this paper, we focus on one of these topics, namely the EU digital COVID certificate (also known as Green Pass), with the objective of uncovering the stance towards it in a specific time period for the Italian Twitter community. To this aim, we first tested some classifiers for determining the most suitable one in terms of performance and complexity for the stance detection problem under consideration. Then, we compared several approaches aimed at counteracting the occurrence of concept drift, i.e., that phenomenon for which the characteristics of the dataset vary over time, possibly resulting in a degradation of classification accuracy. Our experimental analysis suggests that updating the classifier during the stance monitoring campaign is crucial for maintaining a satisfactory level of performance. Finally, we deployed our system to monitor the stance on the topic of Green Pass expressed in tweets published from July to December 2021 and to obtain insights about its
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/11568/1141976
 Attenzione

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

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