16 FEB 2018

Sanremo and Big Data: predictive analytics guesses the winners

Sanremo is the Italian Song Festival which takes place in February in the city of Sanremo, in northern Italy.

If we told you that there is a relationship between Sanremo and Big Data, would you believe it? It seems strange, but it is true. In fact, the conversations on social networks, blogs news sites, and video comments made during the festival were analyzed and highlighted the preferences of viewers regarding the songs, the guests and the most appreciated presenters of the Italian song festival.

Sanremo and predictive analytics

During the festival, the cognitive system of IBM “Watson” analyzed over a million documents to draw up a list of the winners of the 68th edition of Sanremo and the results of this predictive analytics accurately guessed the result. Last year, Watson tried the same analysis, but the winner was not Fiorella Mannoia as expected, but Francesco Gabbani. In addition, Watson “guessed” the winner of “Sanremo Giovani” (a category of Sanremo composed of young singers).

So, this year IBM Watson tried again and during the five festival nights, more than 1 million and 700,000 social documents were analyzed, of which 50,000 were produced only during the 30 minutes of televoting. During this time, an average of 1,760 comments per minute were produced.

From the first night viewers’ favorites, as could be read from the comments analyzed, were Ermal Meta and Fabrizio Moro (2 singers who sang together), The Kolors (an Italian band), Annalisa and Lo Stato Sociale (an Italian band) and on February 10th, the winners were awarded and they were, instead, Ermal Meta and Fabrizio Moro, Lo Stato Sociale and Annalisa (who won the first, second and third place respectively). In addition the predictive analytics announced and guessed the winner of the “Nuove proposte”, Ultimo.

Other preferences emerged from the predictive analytics

The myriad of comments analyzed highlighted other information: which guests and presenters the audience preferred, general appreciation for the festival, the duets and the most appreciated ‘look’. The favorite guest was Fiorello (an Italian comedian), but the guys from “Il Volo” (an Italian operatic pop trio) and Pippo Baudo (an Italian presenter) also received very positive comments. Claudio Baglioni was the best presenter, followed by Pierfrancesco Favino and Michelle Hunziker. Sanremo was excellent this year, according to 65% of comments analyzed. Finally, the look and style considered most elegant were those of Ermal Metal, Fabrizio Moro and Ornella Vanoni.

The power of predictive analytics

These analyses highlighted several very important elements:

  • the huge amount of data that we produce daily and which is increasing day by day, as we have previously pointed out several times in our blog;
  • the importance of analyzing this data to try to understand users’ preferences;
  • how technology is making great steps forward in the development of software that while analyzing data, including data written in “natural language”, is able to return more and more precise results.

Elaborated by Lucia D’Adamo, supervised by Marco Pirrone