Determining Trustworthyness in News
Project by Polygence alum Evan
Project's result
The final outcome of my project was a research paper detailing the process and results of creating a neural network to classify credible and fake news articles.
They started it from zero. Are you ready to level up with us?
Summary
Abstract - Due to the rapid pace misinformation spreads over the internet, it has become a major social issue, negatively influencing public opinion, politics and economics, yet traditional fact-checking is too time consuming to be widely adopted. This study approaches the problem of automating this process with deep learning and semantic parsing. A neural network trained on the ISOT dataset was able to differentiate credible and misinforming news articles with 99.91% accuracy. Semantic parsing revealed that fake news articles exhibit significantly higher negative sentiment intensity scores compared to credible news articles. However, a second article trained on the LIAR dataset was only able to differentiate true and false statements with a 61.62% accuracy due to dataset limitations; this is consistent with a past study on this dataset. The results of this study highlight the potential of deep learning in fake news detection.
Mariel
Polygence mentor
PhD Doctor of Philosophy
Subjects
Literature, Computer Science, Arts, Quantitative
Expertise
Machine Learning, Artificial Intelligence, Computer Science, and Math
Check out their profile
Evan
Student
Graduation Year
2025
Project review
“The project was very engaging and I learned a lot more from this experience than I originally anticipated. I've gained a much better understanding of how machine learning systems work by creating my own hands-on project. My mentor was very helpful and flexible throughout this whole process which improved the overall experience.”
About my mentor
“Mariel was very helpful for me and was able to help me solve almost all the problems I had in my coding and research process. I think her friendly and flexible attitude also contributed to a positive experience at the Polygence program.”