An algorithm analyzed months of tweets about coronaviruses and vaccines, identifying the categories that share the most reliable news and those that divulge the most disinformation.
Social networks like Facebook and Twitter have long started to fight disinformation by adopting different techniques and approaches. Despite the presence of collaborations with fact-checkers, however, fake news always find a way to be disseminated to large audiences on online platforms. But what is the category of profiles that tend to share the most fake news? Thanks to an analysis conducted through an algorithm by Iconsulting, it turns out that in Italy it is the politicians who share the most hoaxes, closely followed by activists and give it entrepreneurs. The percentage of reliability of the tweets published by Italian politicians is equal to 56%, and the figure refers to approximately 1042 tweets and 14,161 retweets.
A worrying revelation, also in light of the fact that during the initial period of the pandemic and with the start of the vaccination campaign, fake news has undergone a significant surge. Confirming this trend are the reports of the Postal Police and the Annual Report on Information Security Policy, which report a 436% increase in fake news compared to the previous year.
Furthermore, with the arrival of the anti-Covid vaccines, a real infodemic of imaginative assumptions, plots and contradictory news has occurred online. A dangerous phenomenon, that of the spread of fake news on vaccines, which has even prompted the Ministry of Health to dedicate a page of its website to their timely denial.
Iconsulting carried out an analysis on the tweets published by the Italian profiles between 27 December 2020 and 19 April 2021, with a particular focus on those who used the hashtags vaccino, vaccini, novax, astrazeneca, PfizerBioNTech, Moderna and Sputnik. The purpose of the Italian data-driven transformation company was to use an algorithm created for the occasion in order to find out which was the category that most of all shared false news.
In addition to having condemned politicians, the algorithm identified that the most trusted category is that of governmental bodies (with a 99% reliability) that throughout this pandemic period have been one of the most reliable sources of information. Obviously if we are talking about viruses, diseases and vaccinations, the profiles belonging to the world of health they proved to be a valuable item, with a reliability index of 91%. The same can be said of traditional media, which – despite the vulgarity – with 90% reliability have proved to be source of authoritative, reliable and verified information.
An algorithm for finding disinformation
“To design actions to combat disinformation, we used an approach based on the phenomenon of the so-called echo-chambers and on the analysis of the correlations between users who share unreliable content.”, Explained a Wired a spokesperson for Iconsulting. Therefore, taking advantage of the machine learning the algorithm was able to learn to recognize, thanks to an initial data set, whether a news was true or false, then assigning a score from 0 to 1 to the tweets analyzed. At this point, the score can also be attributed to users based on how they interact with the tweets examined.
This was able to ascertain that i satirical profiles– cartoonists, authors and bloggers – are much more reliable on Twitter than politicians’ profiles. The reliability index of the world of Italian satire, which seems to feel the responsibility of an important role by committing itself to spreading reliable news, is equal to 80% against a very low 56% of politicians. An example of how satire has managed to convey reliable information is Andrea Lorenzon with its Cartoni Morti channel.
Another fact that emerges from the analysis of the algorithm is that the category of entrepreneurs, who most of all in this period should support the restart of the country by focusing on the vaccination campaign, recorded a decline in reliability of the news tweeted going from 65% to 57%.
In all this, Twitter continues its internal fight against fake news by labeling misinforming news where possible and inviting users to read an article before sharing it again without even opening it. These tricks, which still allow users to have an overall picture of the information regardless of their veracity, have allowed the social network to decrease the sharing of fake news in a delicate moment where it is enough for a word like “thrombosis” to reach uninformed ears to bring down trust in a life-saving vaccine.