"MIning" the sentiment during Covid19 pandemic

Influenced by the health emergency caused by the Covid-19 pandemic, many brands have changed their communication strategy, introducing more or less explicit references to principles of solidarity and fraternity in their TV commercial, to enhance trust and hope in Italian families during the lockdown. The traditional settings of the advertising format, focused on the product quality, have been moved to the background, in favour of the strengthening of the "brand image" by means of specific words, characters, hashtags and music, spreading empathetic messages to all those who need to regain hope and confidence, in a period of extreme emotional fragility. In this scenario, the paper aims to detect the emotions and the awareness raised by the brands during the lockdown period, compared to external periods, and to measure, through Text mining techniques, the sentiment of customers expressed on the social network Twitter. After a careful phase of extraction and pre-processing of tweets by means of Natural Language Processing algorithms, a dataset of 20,982 tweets was analysed, considering three different periods: pre-lockdown, during lockdown and post lockdown. The frequency analysis through the TF-IDF algorithm, firstly highlighted not only an increase in activity on the Twitter social network during the lockdown period of 3 times and half higher than the previous period, but, above all, a completely different use of the terms ‘Italia’, ‘Grazie’ and ‘Nuovo’, the most used in tweets, along with the hashtags '#iorestoacasa' and '#andràtuttobene’. The novelty in the advertising approach of the brands towards consumers, therefore, seems to have generated a certain sense of appreciation and gratitude, as well as a strong sense of belonging that is absent in the two other periods, in which it emerged that the TV commercial was commented on Twitter mostly for the brand's product or as an element of "disturbance" for other television events. The analyses of bi-grams and of co-occurrences highlighted that the commercials mostly drawing consumers’ interest were Lavazza and Barilla. In general, three reversals characterized the trend of the average daily “sentiment score”: a drop in “sentiment” just before the lockdown period, a strong resumption of positivity throughout it and a decline in the immediately following period. These results show how, by moving away from the traditional advertising approach, brands have managed, in order to enhance the amount of people talking about them, to reach the difficult goal of arousing an important positive mood in an extremely emotionally draining period; following, the trend has been reversed probably due to fatigue, and the first tastes of freedom.