Work

Networks, Language and Psychological Biases: how online social media can be manipulated to spread false and partisan information

Type: dissertation
Abstract: Online disinformation can inflict severe damage to the politics, public health and economy of a country.Since the advent of social networks, the distance between producers and consumers of news has shrunk: nowadays anyone can write news that will be read by thousands, thus misinformation can be spread with ease.This phenomenon has been of concern for governments and international organisations, and in the last few years also the scientific community acknowledged the manipulation of online information as a huge threat, to be understood and addressed.Especially from 2016 on, researchers published thousands of articles on the issue.This field of research is young and highly multidisciplinary, and the multitude of perspectives results in immature terminology and in a variety of findings that are often contradictory.We tackle the problem of disinformation by three different perspectives.First, there is the layer of psychological individual and social biases, that affect the individuals' judgement of the information they receive.Biases can make us believe in some blatantly false news if it resonates with our prior beliefs, and if it is perceived as aligned to our social context it will motivate us to further propagate it.Second, we investigate the role of networks as the transmission layer of disinformation.Networks' structure and dynamics play a major role in the diffusion of falsehoods, fostering the emergence of clusters of like-minded people that reinforce each other's beliefs.Last, there are language and emotions, the rhetorical instruments through which malicious actors can deceive the recipients of false news.Negative emotions and a malicious semantic framing of specific concepts can shape the readers' minds, preventing them from an objective assessment of incoming news.These three layers can help us deconstruct the reasons and the outcomes of disinformation spreading, ultimately disentangling a jungle of interconnected phenomena that cause fake news to spread.This three-fold perspective on the issue of fake news is motivated by a data-driven analysis of the thousands of scholarly publications produced in the last few years.For such a reason, we built FakenewsResearch, a selfupdating online repository of scientific articles that helped us define the discipline and its main research areas.Assisted by this tool, we could sort most part of this massive body of publications in the three distinct branches of Psychology, Networks and Language.We then showcase our contributions to the three fields.We provide insights into how users perceive online news as it appears on social networks, and how this perception can be influenced by contextual information that exploits their social biases and prior beliefs.We point out how the source of news and other people's opinions can be two-edged swords, and also how young adults -the population segment more expert on Web and social media -is less vulnerable and keener to fact-checking news.Then, we show the results of an extensive analysis on the activity of malicious actors and automated accounts on social networks, within the context of the divisive debate around immigration, where users divide in clusters depending on their stance towards the topic.These analyses target the spread of unreliable content in a landscape of opinion fragmentation, challenging the common idea of a binary polarisation of opinion.Lastly, we investigate how mainstream and low-quality press framed the COVID-19 vaccination campaign, through the semantic and emotional analysis of their news articles' titles and bodies.We found that mainstream newspapers portrayed vaccines with consistently higher levels of hope, while disinformation outlets targeted some specific concepts -especially the AstraZeneca vaccine around its temporary suspension on March 2021 -with sadness or fear.Together, our studies focus on why people share disinformation, how it spreads, and how it can deceive others.We focus on Italy, while many other studies in literature are focused on the United States.Our works shed light on the disinformation landscape during two critical moments such as the 2019 Italian government crisis and the 2020-21 global pandemic, when politics and public health were especially vulnerable to manipulation.Finally, our contributions bring to the community practical tools such as Fakenews-Research and the two Python libraries PyPlutchik and EmoLib.
Institution: University of Turin
Language: English
Open Access status: green
OpenAlex ID: W7218273134
Cites: 232
Cited by:
Related to: 10