How to mitigate the spread of fake news?

It’s easy to get distracted by the urge to hit ‘share’. But most of us wouldn’t, if we gave it a moment’s thought. A simple intervention is sufficient to make people realize the story is suspect. DeepCheck provides the most simple detection tool for spotting false information and mitigate the arbitrary spread of fake news.

Falling for Fake News

An unprecedented crisis like the COVID-19 pandemic is also characterized by a loss of public trust in various institutions. The spread and dissemination of fake news via the internet and social media channels leads to uncertainty about the veracity and credibility of information available. Most people do not question information and claims, whether they are satire or malicious hoax, sufficiently or at all (Karner, Fintham, 2018). Due to digitalization, news and articles published in the web and social media channels are spreading 10 times faster than real news. Especially regarding important events such as elections with global attention, fake news and false conspiracies are deliberately published in order to gain an undue advantage. Thus a simple tool is necessary in order to mitigate the spread of fake news.

The belief in conspiracies

The novel coronavirus pandemic has given rise to numerous conspiracy theories e.g. about the origins of the virus and whom it infects, and even about the virus itself. The reason people believe in conspiracies can vary widely but they most likely flourish in times of crisis and tend to surround major and unprecedented events that require satisfying explanations. For complex and big issues people want to have simple explanations and comprehensive information. Since a pandemic on this scale has not been seen before there is no definitive roadmap. Almost every country got to come up with their own way of dealing with it. During these times of insecurity the most vulnerable people that are likely to be in danger from the virus, minorities and people with less education are more likely to believe in conspiracies. 

Another major, more complicated factor in the spread of false information and beliefs is in the role of politicians who are downplaying and undermining the virus and its spread consistently. False or unsubstantiated claims lead to a widespread uncertainty among the population and suggest that the pandemic has not been taken seriously. Many claims and hunches are published in social media channels. Since a vast number of people inform themselves through social media channels misinformation can make a huge impact on decisions about life or death. 

How does DeepCheck work?

DeepCheck detects fake news by extracting linguistic characteristics from a large body of news articles. It takes a piece of text and identifies the frequency of specific words commonly used in fake news articles and compares it on the similarity to real news items that it has analyzed before.
Based on the dataset containing many different variables e.g. all caps headings, font-size, special characters. Try it out!

The main challenge, however, is to build a system that can handle the vast variety of news topics and the quick change of headlines online, because computer algorithms learn from samples and if these samples are not sufficiently representative of online news, the model’s prediction would not be reliable. 

For a further development and improvement of the algorithm in order to be able to distinguish with confidence between previously unseen real or fake news articles and white papers, one option is to have human experts collecting and labeling a sufficient quantity of fake and real news articles about COVID-19. With this dataset our machine learning algorithm will be able to find common features that keep occuring in each collection regardless of other varieties. 

Genuine news vs fake news

Genuine news and sophisticated science journals about COVID-19, on the other hand, contain a larger proportion of words related to scientific language and expressions covering virology and epidemic science. Usually sources on each critical statement and claim are implemented or linked with a dofollow link in the written text. This could be an indicator for genuine news and are worth considering in the development process of the algorithm.

Spotting and mitigating fake news

DeepCheck identifies linguistic characteristics to spot fake news using machine learning and natural language processing technology. Our research on a large collection of fact-checked news articles, science journals and white papers on a variety of topics regarding COVID-19 shows that, on average, fake news articles use more words related to sex, death and anxiety. Many expressions used are also common in hate speech. It can be assumed that a linguistic and stylistic approach combined with machine learning may be useful in spotting misleading information and suspicious news. 

Recurring writing style and linguistic

Machine learning algorithms are able to instantaneously complete tasks that would have taken a human much longer. When machine learning is applied to natural language processing it allows us to build text classification systems that distinguish several articles from one another.

Recent scientific research helped us to understand the characteristics of fake news in order to develop a technology to help readers detect misinformation and thereby mitigate the spread of fake news.

One approach examines the writing style of an article rather than its published origin. By analyzing linguistic characteristics of a written text we are able to understand the motives of the authors. With DeepCheck we can detect whether specific words and phrases tend to occur more frequently in a deceptive article compared to one written and published genuinely. 

Rated Author
Share this blog post!
Share on facebook
Share on twitter
Share on linkedin
Search other blog posts!
Support DeepCheck