How science and technology can fight fake news

It is undeniable that fake news became a big contemporary issue. What to believe, and what instead is a lie? I remember myself at high school thinking of how lucky my generation was, having the possibility to access a huge amount of ready-available news. Centuries ago, when news were brought through villages by men riding horses, people learnt about important events with big delays, and often information was distorted. This, together with ignorance of people, allowed their opinion to be easily manipulated. Paradoxically, nowadays we are living an opposite situation, but the outcome is the same. The burden of fake news is able to distort people opinion about nearly any topic, and issues that have been obvious for decades are now questioned. Are vaccines really safe? Did evolution really happen? Is climate really changing? And so on. Any topic can be contaminated by fake news, although politics and science are paying the highest price.

Let’s see how belief in science is challenged by fake news.

Why do people lack belief in science?

Vaccines, evolution and climate change are three main topics targeted by fake news. From where does this disinformation come from? First, a lexical preamble should be done: please pay attention to the word: “disinformation” differs from “misinformation” in the feature that the former has the intention to distort the truth, while the latter can be accidental. Instead, the word “uninformed” refers to someone who does not have information about a certain topic. This is a typical situation for scientific facts: people are uninformed about how a process occurs, therefore they are easy to disinform. It is government’s responsibility to provide correct and easily understandable information to everyone whenever a new technology becomes available, or whenever a topic fosters fear in the population.

Distrust in science is a major issue, as demonstrated by US Science & Engineering Indicators (SEI) surveys: lack of people’s knowledge in science is massive, but it is also relevant that 44% of Americans think this as an issue to solve. What is most shocking for a scientist (like me) is that Americans have little trust in scientists. In 2014, 67% of people interviewed believed that scientists have no clue about GMO effects, 52% that scientists have different opinions about the Big Bang theory (no, not the TV series), 37% that they disagree with climate change and 29% with evolution. Well, the scientific community doesn’t have any doubts about any of these topics. GMO effects are well known and documented, Big Bang, climate change and evolution are irrefutable real facts.

Another cause of mistrust in science is due to people’s belief in conspiracy theories. Usually, these theories refuse science as an explanatory tool for phenomena, because it is too… obvious. Uninformed people are easy prey for the disinformation machinery, especially since when these theories are supported by politicians and other relevant people (musicians, actors).

Why are fake news so spread and difficult to fight?

There are several points to consider about this. First, the inability to recognize disinformation (or misinformation). In other words, people are not able to evaluate critically an information, and this is in line with the absence of education to this purpose, especially for older generations. Websites were created to support people in debunking fake news, like www.factcheck.org/ and www.politifact.com, and specific algorithms to fight disinformation have been developed by social media. Often, not even this effort is enough, because people’s psychology fights against it. People prefer news that follow a logical flow, believing the source they think it’s true. This means that they will accept only news supporting their beliefs and ideas, while rejecting everything against them. Emotion also plays a crucial role in this, meaning that we choose to believe a news that touches our feelings.

At a society-level, a fake news gains importance proportionally to the frequency we learn about it. If the first time we feel uncertain if to believe it or not, after we meet it ten times we don’t doubt anymore. This generates a cascade mechanism, because the more people read and share a news, the more will believe it and contribute to its spreading. Politics plays a major role in this, because politicians have interest in shaping people’s minds according to their ideas.

How to fight fake news?

Initially, the danger of fake news was underestimated, but quickly it became evident that it is a major issue to fight. As we just saw, people are not very good at spotting disinformation. The International Federation of Library Associations and Institutions (IFLA) and Associated Press, in 2017, developed some frameworks to help people in spotting fake news, but they are quite laborious to follow. Last year, Choy and Chong designed an easy framework, called LeSiE, able to identify disinformation by exploiting three main features of fake news, which distinguish them from normal articles:

  1. Lexical structure: fake news use a specific language, characterized by active verbs, hedges, suspenseful language and abuse of numerals.
  2. Simplicity: fake news never tell complex stories, preferring short sentences and short words.
  3. Emotions: fake news need to attract attention, therefore they always display a “viral” content rather than a normal one.

 Authors proved that LeSiE can identify a fake news just by looking at:

  1. Strong positive or negative words.
  2. Length of the title in terms of number of letters.
  3. Preponderance of verbs, adjectives, names or numbers.

A new step in fighting fake news is provided by computers. Artificial intelligence is progressing very fast (read here), and computers can be trained to identify fake news in the World Wide Web. How? By following the aforementioned frameworks, and computationally estimating if a news can be reliable or not.

Other strategies to spot fake news exploit connections between words within sentences, while others focus on the news’ spreading process.

Regarding the first strategy, it is worth mentioning Giovanni Luca Ciampaglia’s recent work. Together with his colleagues at the Indiana University in Bloomington, he developed an algorithm able to analyze simple sentences (subject, predicate and object) and estimate, according to the “noun network” the virtual distance between the words within the sentence. The noun network exploits DBpedia, a tool that collects all structured contents from Wikipedia. To clarify how DBpedia works, imagine to take a word, let’s say “spouse”. DBpedia will return a network of nouns related to the chosen one. For “spouse”, the most closely related words are “child”, “relative” and “parent”. A bit farer but still related, we find “relation”, “starring partner” and “guest”, together with many others. If, in our sentence, the “knowledge stream” between subject and object is too long (meaning that there is not a close interaction between the two words according to DBpedia), there is an indication that the sentence could be misleading.

Another feature that distinguishes fake from legitimate news and that can be used by artificial intelligence, is the way of spreading. Usually, legitimate news have a single source, while fake ones spread through people who share it, in a chained and sparser way. This was highlighted by Zhao and colleagues, who analyzed spreading of real and fake news on Twitter in Japan and on its Chinese counterpart Weibo. They show that fake news spread in a multiple broadcast-like way, without a dominant source, which instead is prominent in legitimate news.

References

Dietram A. Scheufele, Nicole M. Krause. Science audiences, misinformation, and fake news. Proceedings of the National Academy of Sciences Jan 2019, 201805871; DOI:10.1073/pnas.1805871115

Murphy Choy, Mark Chong. Seeing Through Misinformation: A Framework for Identifying Fake Online News.
arXiv:1804.03508

Prashant Shiralkar, Alessandro Flammini, Filippo Menczer, Giovanni Luca Ciampaglia. Finding Streams in Knowledge Graphs to Support Fact Checking. arXiv:1708.07239

Zilong Zhao, Jichang Zhao, Yukie Sano, Orr Levy, Hideki Takayasu, Misako Takayasu, Daqing Li, Shlomo Havlin. Fake news propagate differently from real news even at early stages of spreading. arXiv:1803.03443

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s