Abstract

With the advent of modern technology, primarily based on Artificial Intelligence, the amount of information available has increased exponentially compared to the past. Among the various drawbacks, we can also identify the non-reliability of many sources of information. Social Media has given a strong impetus to publishing content, thanks also to the low cost related to searching and sharing information. Consequently, the link between the source and the value of using the information has been disintermediated, increasing the scale and the speed of its spread, but lowering the credibility of the source which is rarely verified. In this work, we analyze some metrics proposed in the literature for the measurement of the phenomenon of opinion polarization and, as a case study, we apply these metrics to a dataset containing the results of an American survey that collected the opinions of users on various issues concerning social issues. The application of the various metrics made it possible to identify a certain degree of polarization in the dataset that can be mapped on the two main American political alignments. Finally, given the correlation between the various variables involved in the analysis, we have developed a neural network capable of predicting the user’s political alignment with very high accuracy