La semiotica può migliorare l’apprendimento supervisionato delle reti neurali? Il caso di studio del tweet di Papa Francesco

The paper focuses on a case–study: a corpus of 1234 Italian tweets in reply to Pope Francis’ ecological tweets related to the encyclical letter Laudato si’ has been collected and labelled by the research team of the Semiotic and big data lab at the University of Turin using semiotic categories to substitute the vague notion of “subjectivity” in use in sentiment analysis. A simple neural network has been trained on the corpus to classify messages into “history” and “discourse” with a final accuracy score of 97%. The paper explores the practical and social implications of the feasibility study as well as its limitations and suggests further transdisciplinary research using semiotics as a standard vocabulary to increase cooperation between social and computer sciences. The analysis is accompanied by a historical premise and a brief analysis of the concordance between Saint Francis of Assisi’s Canticle of the Creatures and Pope Francis’ encyclical letter Laudato si’.