Soon professionals and voters alike could have the power to fact check politicians’ claims as soon as they make them.
Chengkai Li, principle investigator and associate computer science professor, and his team of graduate students are working to make that dream a reality with the help of the problem-solving computer program ClaimBuster.
ClaimBuster analyzes the closed captions from political debates and ranks each claim made by politicians by what has a higher likeliness of needing to be checked, according to their website.
Claims deemed more check-worthy are claims that may need to be checked by professional fact checkers.
Those less check worthy are less likely of needing to be checked. Each sentence is scaled by how check-worthy it is by color and number, leaving fact checkers to do the rest, said Naeemul Hassan, graduate teaching assistant, and research and development leader for the project.
The program learns from training data, Li said. People participate in data collection by filling out surveys of what they deem worthy of fact checking and importance.
The program then ranks the check-worthiness of politician’s claims based on the training data from those who filled out surveys.
Number amounts and changes in verb tenses tend to lead to a higher check-worthiness score, Li said.
The project’s intention is to be helpful to journalists by fact checking claims likely to be important to the public, according to the Knight Foundation’s ClaimBuster website.
Instead of focusing on claims not as check-worthy or worth reporting on, the program would help journalists find claims people would be interested in.
Assistant communication professor Mark Tremayne is the co-principle investigator for the project.
Tremayne discussed using the debates for the project from the beginning, Li said.
“This is a computational journalism project,” Hassan said. “[Tremayne] is helping us from the journalism perspective.”
In November, the project was awarded a $35,000 grant by the Knight Prototype Fund.
Researchers are given six months to prepare and conduct further research before building, according to their website.
The project also received assistance from the National Science Foundation Innovation Corps in the form of a $50,000 grant.
By April, the team wants to display a more refined prototype website with additional capabilities and features, Hassan said.
Goals include more social media features, such as discussion boards, and getting the project to cover live events.
The debates currently featured on their prototype are previously recorded to show proof of the concept.
Gensheng Zhang, computer science doctoral student, and Hassan served as the entrepreneurial leads in the commercializing part of the research project and interviewed 100 customers over seven weeks to assess if ClaimBuster would meet the needs of businesses, Li said.
Fatma Dogan, computer science doctoral student, who has been working on the project since she was a graduate student, is leading ClaimBuster’s Twitter component.
If any politician’s tweet is deemed check-worthy, it is retweeted to the account, she said.
Li said making fact checking more available to everyday people could help keep politicians in check in regards to the claims they make.
Li and his team have turned to crowdsourcing to complete more than 20,000 claims needed by the program to continue its training.
To reach this goal, surveys need to be filled out, Li said.
Students can help with data collection for the program by registering on ClaimBuster’s website and filling out surveys of what they deem worthy of fact checking.