Filippo Menczer, a professor of computer science and informatics at the School of Informatics, Computing, and Engineering, has been awarded a grant from Craig Newmark Philanthropies, a philanthropic organization endowed by the founder of Craigslist, to fund two projects that will help journalists curb the spread of online misinformation.
One effort looks to develop Hoaxy, a web tool that tracks how misinformation spreads on Twitter by monitoring sources that typically post various kinds of misleading or biased information. The project aims to make Hoaxy more usable for journalists and the general public, and to possibly integrate it into other tools used in newsrooms. The second project will study the feasibility of computational fact checking methods to match claims against knowledge graphs similar to those used by search engines to understand queries.
“Craig Newmark is a philanthropist who supports many efforts to study misinformation and to promote a healthy news ecosystem,” Menczer said. “These projects are critical to understanding where claims originate, and who is spreading misinformation, conspiracy theories, hoaxes, and biased information. His organization is very generous and has funded both of these efforts.”
Menczer and his group unveiled Hoaxy earlier this year, and his Botometer tool uses machine learning to judge whether a Twitter user is likely to be a bot or a human. The combination of Botometer and Hoaxy has led to the development of Bot Electioneering Volume, a tool that measures and visualizes electoral activity by social bots.
The use of social media in the spread of misinformation has blurred the line between fact and fiction, and Menczer’s computational fact checking could help bring that line back into focus.
“For instance, you might have the claim ‘Obama is a Muslim,’ ” Menczer said. “Barack Obama is an entity that exists in most knowledge graphs. ‘Muslim,’ a description of people who follow the Islamic religion, also exists in knowledge graphs. The ‘is a’ represents a relationship that says one object is a member of a category. It’s the same concept that allows a search engine to give you structured results, such as when you search for ‘Paris,’ and you get information about the city, maps, and so on. Our project wants to use this kind of semantic network representation to check whether there is a link between entities mentioned in a claim, and whether it is consistent with the claim.”
By studying links to key phrases and whether they come from trusted sources or ones without a solid reputation, Menczer hopes the foundation can be laid to alert users to the veracity of claims being made online.
“Very often, claims are copied,” Menczer said. “If people see that similar claims are not coming from reputable sources, they might be able to make a more informed decision about whether a claim is accurate.”
The awards are worth a total of $156,000 and will involve SICE students, assistant professor Giovanni Luca Ciampaglia from the University of South Florida, and researchers from the Indiana University Network Science Institute (IUNI) and from Harvard University’s First Draft.
“The spread of misinformation is an important issue that impacts our daily lives, and Fil’s efforts to identify and combat the problem have been critical,” said Raj Acharya, dean of SICE. “The support of Craig Newmark Philanthropies will allow SICE to continue to play an important role in the battle to lessen the impact non-human actors can have on shaping our general discourse.”
For more information on the study of complex networks and systems at SICE, visit our website.