New tool to analyze the spread of 'fake news' on social networks

The spread of ideas on social media is an increasingly important phenomenon, which plays a crucial role in understanding how fake news proliferates, as well as its impact on democratic processes.

Situations such as the covid-19 pandemic, in which there have been many occasions in which social networks have been used to disseminate non-verified information, even relating the disease to 5G networks, highlight the importance of account of the role that social networks play in the dissemination of information and, above all, of disinformation.

From WhatsApp groups of students, it has been observed that the presence of only 1% of critical individuals or bots can reduce the time for a news item to reach half of the population connected to a social network by 20%.

Now, a work developed by the researchers Jesús San Martín of the Polytechnic University of Madrid (UPM), Fátima Drubi of the University of Oviedo, and Daniel Rodríguez Pérez of the UNED, has used a  mathematical model to analyze the possibilities of a rumor being transmitted more or less quickly in a social network and the influence that the polarization of individuals has on this process. The study is published in the journal  Mathematics and Computers in Simulation.

This story also appeared in SINC

“The spread of fake news through social networks and its impact on today's society is more than evident. The US electoral campaign or the Brexit referendum, both in 2016, as well as the biased news campaigns detected in Catalonia in 2018, are just some examples of how these processes are reconfiguring societies and affecting democracies”, explains Jesús San Martín, from the Higher Technical School of Engineering and Industrial Design of the UPM. 

"Having tools such as the one we have developed to infer the mechanisms of propagation of this news and differentiate the" rumors of a lifetime "from the fake news intentionally propagated to attack our society is, now more than ever, of great social importance".

Three parameters of the social network

Taking into account this difference based on the intentionality of spreading false news, the authors simulated the spread of a rumor based on three characteristic parameters of a social network: the probability that an individual knows the initial rumor, the probability of that a non-polarized individual who receives the news share it with their contacts and their groups, and the proportion of the population made up of polarized individuals (uncritical of the content of the news) or bots  (a computer program that automatically performs repetitive tasks) that propagate the rumor automatically as soon as it reaches them. 

This analytical tool allows inferring what is happening in a social network from how a rumor evolves in it and the behavior of individuals in relation to its transmission.

“Our objective was twofold. On the one hand, we wanted to find the law that governs the evolution of the spread of a rumor in the network and to find out how long this rumor will reach a given fraction of the individuals connected to that network. On the other hand, it seemed essential to us to detect the presence of groups of bots or uncritical individuals, who automatically forward a certain rumor in a coordinated way, and to see how they affect the spread of the news, "says the researcher. 

“Taking as a starting point a social network model whose structure we derive from the WhatsApp groups of students, the results showed that the presence of only 1% of bots or uncritical individuals can decrease by 20% over time. necessary for a news item to reach half the population connected to a social network”, he adds.

The researchers also used their model to fit published empirical data on the spread of hoaxes on WhatsApp. In the case analyzed, its model predicts that a news item would reach half the population in less than 6 days, and would reach 99% of the network in 3 and a half months. 

For the team, the importance of these results lies in the fact that they provide an analytical tool that allows inferring what is happening in a social network from how a rumor evolves in it and the behavior of individuals in relation to its transmission.