Interview: How fake news spreads on social media

Northern Kentucky University has created a new endowed professorship of computer science. The STRAWS Endowed Professorship of Computer Science was established with the support of Dr. Carol Swarts, a physician and longtime friend of the university. It is the first endowed faculty position awarded to a woman in the university's College of Informatics.

Earlier this month, NKU and Dr. Swarts announced Alina Campan as the inaugural STRAWS Professor of Computer Science.

Campan's research on data mining and data privacy earned a Yahoo Research Best Paper award in 2008. She collaborates with faculty in NKU’s journalism, computer science and statistics departments to understand the spreading of disinformation on social media.

NKY thrives Managing Editor David Holthaus interviewed Campan about social media and disinformation. She was joined by her colleague, Marius Truta, an NKU professor of computer science.

How have Facebook and Twitter become such vehicles for disinformation?

Alina Campan: The main thing is that everyone can post whatever they want, whenever they want, so it's not content that goes through a review before it’s published. It’s the freedom of pushing out your information and your thoughts in a matter of seconds. Before, we had newspapers, TV, and radio, but with these platforms, everyone can within seconds post whatever they think.

Should Facebook and Twitter do more to evaluate the information on their platforms?

Campan: I think they’re trying, because it’s a matter of reputation. People tend to not trust platforms that have a bad reputation. But it’s hard because you can’t rely so much on human decisions because of the huge amount of information to check, and that’s the most reliable way to check information. The automatic ways, no matter how evolved they are, they’re not going to be as good as humans are. They’re not at the point where they can decide on the truthfulness of information in automatic ways.

Marius Truta: When you deliver information in any format, there may be some that is true and some that is false. It’s never 100 percent false. That’s where the problem is. Facebook and Twitter can’t be the guardians of truth. Once you start censoring the information because it’s not all true, you’ll have people complaining.

Campan: It depends on what kind of information you have. If it’s factual, you can check it. But once you start giving your opinion on something, it’s much different.

Truta: We actually wrote a paper on this three years ago. We came to the conclusion that computer science researchers should not get involved in this issue of what is true and what is false.

What are some of the key methods used to spread disinformation on social media?

Campan: Many of them start by building a website with fake information but it looks pretty legitimate, so it's not too different from legitimate websites. The attackers will create a number of accounts and then they infiltrate communities they believe are susceptible to believe the fake news. The members of the communities then connect with them, either by following them or by sending messages. As soon as the original users believe or “like” the website with the fake information, it will grow and spread to other communities, sometimes in an exponential manner.

Has Russia been out front and how have they done that?

Truta: Spreading propaganda has existed for a long time in history. Actually, the British empire more than one hundred years ago was at the forefront of the propaganda machine. In particular, during the First World War. Other countries tried to catch up. These existed well before social media.

How effective is social media disinformation in a political campaign?

Truta: It might reinforce beliefs, so it might not be as effective in changing opinion. It may also make beliefs more extreme. So the middle ground becomes slimmer. People will go further to the left and the right.

What should people do to evaluate the quality of the information?

Campan: First and foremost, users should get educated. NKU has an information literacy program where they try and teach students how to evaluate information correctly. Is the source reliable? Can you check if the information exists or is it made up? This is like educating people to detect phishing attempts or credit card fraud.

What are some of the automatic methods social media platforms use to flag posts?

Campan: One way is to start with a set of documents that a human has already looked at and attach tags and say this document is fake or this document is legitimate. Machine learning will try to learn patterns to differentiate between fake and true news. Those are based on looking at the content and the other contextual information that comes with those documents. You can make a model that will try to predict based on what it has already learned from these known and tagged documents. That model will try and make predictions on things that it hasn’t seen before.

How reliable is that?

Campan: Some do go up to 80 percent accuracy, but you still have that margin that’s incorrect. And you can have false positives and false negatives.

How do Facebook and twitter differ in their ability to spread disinformation?

Campan: Facebook has more privacy mechanisms built into it. It’s not supposed to be used for spreading information widely. Twitter’s purpose is spreading information, it’s microblogging. You want to spread your opinions as widely as possible.

Truta: Facebook is more grass roots because everything has to be done in smaller groups, smaller communities. Twitter, on the other hand, when the president or Justin Bieber says something, everyone sees it right away. It’s like CBS or NBC News was 50 years ago.

What social media platforms do you see emerging?

Campan: Different age groups are using different platforms. My students say Facebook is old news, they’re not using it anymore. They’re not even using Twitter that much. They’ve moved to Instagram or other platforms. It depends on the user's age. Facebook users are older.

Truta: The problem with Facebook is kids don’t like to have their parents as friends.

Campan: For our technology groups, I see Discord groups being used quite a lot. Students are interested in playing certain games. Tik Tok seems to be a platform they use a lot. It's going to be interesting to see what platforms jump ahead as a result of the pandemic and all the digital media that is being used.

Truta: I always ask my students about their hobbies. What I notice is 80 percent of them are doing games. Nobody is really into social media per se. They’re moving into gaming. This is actually a surprise. But this a computer science community, which is more introverted.

Signup for Email Alerts