20 octobre 2022
https://creativecommons.org/licenses/by-nc-nd/4.0/ , info:eu-repo/semantics/openAccess
Leonardo Ranaldi et al., « KERMIT for Sentiment Analysis in Italian Healthcare Reviews », Accademia University Press, ID : 10.4000/books.aaccademia.11017
In this paper, we describe our approach to the sentiment classification challenge on Italian reviews in the healthcare domain. Firstly, we followed the work of Bacco et al. from which we obtained the dataset. Then, we generated our model called KERMITHC based on KERMIT (Zanzotto et al. 2020). Through an extensive comparative analysis of the results obtained, we showed how the use of syntax can improve performance in terms of both accuracy and F1-score compared to previously proposed models. Finally, we explored the interpretative power of KERMIT-viz to explain the inferences made by neural networks on examples.