Sentiment scores measure the strength of customer sentiment when evaluating a product or service. This score is expressed as positive (and negative) for a numerical value between 0 and 100, where 100 is the most favourable possible result, and 0 is the least. This paper aims to combine a product’s sales volume time series with the sentiment score time series of tweets generated by the BERT-NN within a state space model. We apply this model to the monthly sales volume of the Fiat L500 time series from August 2012 to Dec 2018.
Basili, R., Croce, D., Iezzi, D.f., Monte, R. (2023). The Role of BERT in Neural Network Sentiment Scoring for Time Series Forecast. In A.A. Paola Cerchiello (a cura di), Proceedings of the Statistics and Data Science Conference (pp. 55-60). PAVIA : Pavia University Press.
The Role of BERT in Neural Network Sentiment Scoring for Time Series Forecast
Basili R.Validation
;Croce D.Validation
;Iezzi D. F.
Membro del Collaboration Group
;Monte R.Methodology
2023-05-01
Abstract
Sentiment scores measure the strength of customer sentiment when evaluating a product or service. This score is expressed as positive (and negative) for a numerical value between 0 and 100, where 100 is the most favourable possible result, and 0 is the least. This paper aims to combine a product’s sales volume time series with the sentiment score time series of tweets generated by the BERT-NN within a state space model. We apply this model to the monthly sales volume of the Fiat L500 time series from August 2012 to Dec 2018.| File | Dimensione | Formato | |
|---|---|---|---|
|
PAVIA.pdf
accesso aperto
Licenza:
Non specificato
Dimensione
420.7 kB
Formato
Adobe PDF
|
420.7 kB | Adobe PDF | Visualizza/Apri |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
