In the era of social networks and big data Twitter represents a tremendous and cheap source of data able to provide valuable information about any possible topic. Such a source requires techniques to transform text into sensible numerical indexes. In this paper we consider the daily sentiment score measured by two lexicons (AFINN and Bing) on tweets collected for UK from January 15th to February 15th, 2018. As the analysed daily tweets are geolocated at the NUTS subarea level, we focus on the comparison of the two score distributions across regions and serially. Results show that the two lexicons perform very similarly. However, our analysis shows that the sentiment estimated using the tweets doesn’t correlate with the individual well-being estimated using data from an European survey (ESS).
Nell’epoca dei social network e dei big data Twitter rappresenta una fonte di dati straordinaria ed a basso costo, capace di fornire preziose informazioni riguardo ad ogni possibile argomento. Questa fonte di dati richiede, però, metodi in grado di tradurre il testo in indicatori numerici. In questo lavoro si prendono in considerazione i punteggi stimati attraverso due dizionari (AFINN e Bing) su tweet geolocalizzati a livello di NUTS raccolti per il Regno Unito tra il 15 Gennaio ed il 15 Febbraio 2018. Confrontando le distribuzioni regionali dei punteggi e la loro evoluzione storica, si notano risultati molto simili per i due dizionari. Tuttavia non sembra esserci correlazione tra i livelli rilevati mediante sentiment analysis ed il livello di benessere emerso usando i dati dell’indagine ESS.
(2020). Estimating the UK Sentiment Using Twitter . Retrieved from http://hdl.handle.net/10446/167016
Estimating the UK Sentiment Using Twitter
Toninelli, Daniele;Cameletti, Michela
2020-01-01
Abstract
In the era of social networks and big data Twitter represents a tremendous and cheap source of data able to provide valuable information about any possible topic. Such a source requires techniques to transform text into sensible numerical indexes. In this paper we consider the daily sentiment score measured by two lexicons (AFINN and Bing) on tweets collected for UK from January 15th to February 15th, 2018. As the analysed daily tweets are geolocated at the NUTS subarea level, we focus on the comparison of the two score distributions across regions and serially. Results show that the two lexicons perform very similarly. However, our analysis shows that the sentiment estimated using the tweets doesn’t correlate with the individual well-being estimated using data from an European survey (ESS).File | Dimensione del file | Formato | |
---|---|---|---|
Pearson-SIS-2020-atti-convegno_estratto.pdf
Solo gestori di archivio
Versione:
publisher's version - versione editoriale
Licenza:
Licenza default Aisberg
Dimensione del file
1.18 MB
Formato
Adobe PDF
|
1.18 MB | Adobe PDF | Visualizza/Apri |
Pubblicazioni consigliate
Aisberg ©2008 Servizi bibliotecari, Università degli studi di Bergamo | Terms of use/Condizioni di utilizzo