Presenter · Making Learning Visible (Dec 2025)
Can AI Spot a Lie?
Fake-news classifier trained on 44,000+ labeled articles, with a dashboard of its metrics and the words behind each prediction.
- My role
- Solo project
- Team
- Solo project
- Event
- Making Learning Visible (Dec 2025) · Bellevue College · RISE Learning Institute
- Year
- 2025
The problem
Misinformation spreads through emotional language, sensationalism and missing sources. Can a simple, explainable model catch those patterns?
What I built
- A text pipeline: cleaning, stopword removal and lemmatization with NLTK.
- TF-IDF features (unigrams + bigrams) and a Logistic Regression classifier trained on the Kaggle Fake and Real News dataset.
- A multi-page Streamlit dashboard with accuracy, precision, recall, F1, a confusion matrix and the most influential words per class.
- In-app model retraining from the sidebar.
Results
- Presented at Making Learning Visible, Bellevue College's college-wide showcase (Dec 2025).