The COrona VIrus Disease 19 (COVID-19) pandemic required the work of all global experts to tackle it. Despite the abundance of new studies, privacy laws prevent their dissemination for medical investigations: through clinical de-identification, the Protected Health Information (PHI) contained therein can be anonymized so that medical records can be shared and published. The automation of clinical de-identification through deep learning techniques has proven to be less effective for languages other than English due to the scarcity of data sets. Hence a new Italian de-identification data set has been created from the COVID-19 clinical records made available by the Italian Society of Radiology (SIRM). Therefore, two multi-lingual deep learning systems have been developed for this low-resource language scenario: the objective is to investigate their ability to transfer knowledge between different languages while maintaining the necessary features to correctly perform the Named Entity Recognition task for de-identification. The systems were trained using four different strategies, using both the English Informatics for Integrating Biology & the Bedside (i2b2) 2014 and the new Italian SIRM COVID-19 data sets, then evaluated on the latter. These approaches have demonstrated the effectiveness of cross-lingual transfer learning to de-identify medical records written in a low resource language such as Italian, using one with high resources such as English.

Over the last decade industrial and academic communities have increased their focus on sentiment analysis techniques, especially applied to tweets. State-of-the-art results have been recently achieved using language models trained from scratch on corpora made up exclusively of tweets, in order to better handle the Twitter jargon. This work aims to introduce a different approach for Twitter sentiment analysis based on two steps. Firstly, the tweet jargon, including emojis and emoticons, is transformed into plain text, exploiting procedures that are language-independent or easily applicable to different languages. Secondly, the resulting tweets are classified using the language model BERT, but pre-trained on plain text, instead of tweets, for two reasons: (1) pre-trained models on plain text are easily available in many languages, avoiding resource- and time-consuming model training directly on tweets from scratch; (2) available plain text corpora are larger than tweet-only ones, therefore allowing better performance. A case study describing the application of the approach to Italian is presented, with a comparison with other Italian existing solutions. The results obtained show the effectiveness of the approach and indicate that, thanks to its general basis from a methodological perspective, it can also be promising for other languages.