Italian Institutional Communication in Pandemic Period: A Chronological Analysis of Prime Minister Speeches

Since the beginning of 2020 due to the COVID-19 pandemic, most government authorities around the world have granted precautionary measures to reduce the infections that had an impact on everyone’s life, thereby causing huge economic losses. In this framework, communication, especially that provided by official channels, has played a crucial role. In Italy, the prime minister in office held several speeches and they were broadcast through his official YouTube channel. This paper aims to study Italian presidential speeches during the COVID-19 pandemic, focusing on the topics and expressions used by the official communications. Some points of variation are identified in the time series of contagions associated with four pandemic periods. These periods analyze grouped speeches using an unsupervised approach based on explorative cluster analysis and probabilistic topic model. Finally, the topic classification is used to train two supervised predictive models. A neural network and a support vector machines are trained by document-term table obtained by selecting the most popular topic. Good performances, with accuracy greater than 0.8, are obtained for both models, although the best result is obtained by the neural networks with 0.85 of accuracy.