24/11/2025
'Emotions don't lie': New methods for combating fake news on social media
"Post-truth" is one of the most common terms we encounter in articles describing the current era in recent years. Among the most fundamental elements of this post-truth world is fake news, which became increasingly evident with Donald Trump's statements during the 2016 US presidential election and has accelerated on social media during the pandemic. Unfounded news affects institutions and organizations from all segments of society. Recall the 4 percent drop in Pepsi shares in 2016 due to a false report about its CEO. Or the increasing hostility to vaccines due to the impact of false news during the pandemic. The attempts to burn down 5G towers in Europe following rumors that they "lower our immunity" are also fresh examples in our minds. Fake news poses a major threat to social well-being: It's not for nothing that the World Health Organization (WHO) warns that "misinformation and propaganda spread faster than the pandemic itself."
In the last couple of years, with the emergence of artificial intelligence into our lives and the ever-expanding scope of its capabilities, examples of fact and fiction intertwining have multiplied on social media. Let alone text, we can no longer be certain of the authenticity of even a photo or video appearing on platform X (formerly Twitter). Elon Musk's radical changes not only to the platform's name but also to its usage after its acquisition, along with his efforts to transform citizen journalism into X-based journalism, have significantly increased the amount of fake news. Furthermore, we are living in a period of increasing social polarization, increasing distrust, and the proliferation of conspiracy theories in many areas. The negative impacts of fake news on public trust, freedom of expression, and economic prosperity are clear. Perhaps more than ever, we need methods that will facilitate the work of those seeking accurate news and truth while also supporting the fight against fake news.
A paper co-authored by Nihat Kasap from Sabancı Business School, along with Bahareh Farhoudinia, another PhD graduate from the same institution, and Selcen Öztürkcan from Linnaeus University, can be considered an effort in this direction. The article, titled "Emotions Unveiled," published in May 2024, emphasizes emotional intelligence and sentiment analysis, demonstrating that search models that factor in negative emotions can significantly improve the performance of methods used to detect fake news on social media.
Fake news instills fear, true news inspires confidence
"Emotions Unveiled" is based on a dataset consisting of 10,700 posts and hashtags related to COVID-19 in X, collected in August and September 2020, the year the outbreak was declared a pandemic. Analysis of the data using different lexicons and machine learning models reveals eight distinct emotional states: anger, anticipation, disgust, fear, joy, sadness, surprise, and trust. Of these, anger, disgust, and fear are more prevalent in fake news posts, while anticipation and surprise are more prevalent in posts reflecting real news. The emotion most frequently evoked by both fake and real news is fear. Fear is followed by trust. However, the intensity of these emotions also varies. While fear is more prevalent in fake news than in real news, the opposite is true for trust. While news reflecting reality more often triggers positive emotions like anticipation, surprise, and trust, fake news more often evokes negative emotions like anger, disgust, and fear.
Have we become addicted to bad news?
The constant bombardment of bad news, stemming from numerous crises worldwide, has led to the normalization of many previously taboo topics, leading to desensitization. One consequence of this is "doomscrolling". Social media users, instead of avoiding bad news when exposed to it, continue to scroll down their screens, deliberately drawn into it out of fear of missing out or worrying that they should know about the dangers beforehand.
The results of the "Emotions Unveiled" study also reveal the social and psychological dynamics behind social media users' "bad news addiction." Due to negativity bias, bad news, emotions, and feedback have a stronger impact on people than positive experiences. Similarly, social media users prefer to spread more negative posts. Another finding of the study is that fake news detection models yield better results when supported by sentiment analysis methods.
With its content and findings, which will be particularly useful for experts and researchers in fields such as communication, marketing, psychology, and sociology, the "Emotions Unveiled" study can make significant contributions to studies on social media behavior and fake news detection.
Key findings:
- Fake news erodes public trust, harms freedom of expression, and harms the economy.
- Sentiment analysis, as well as news content and social context, can play a significant role in detecting fake news.
- Social media users prefer to spread more negative posts.
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4833527
Farhoudinia, B., Ozturkcan, S., & Kasap, N. (2024). Emotions unveiled: Detecting COVID-19 fake news on social media. Humanities and Social Sciences Communications, 11, 640.
