Fake News Detector

• 

  this is a interface of facenewsdector

Inspiration Traditional detectors only spot how news is written. We wanted to build a tool that also understands what the news is actually saying by combining local ML with real-time AI.

What it does TruthGuard is a hybrid dashboard that uses Machine Learning to detect sensationalist writing patterns and the Gemini API to fact-check claims against real-world knowledge.

How we built it * Frontend: Streamlit for a clean, interactive UI.Local ML: Scikit-learn (PassiveAggressiveClassifier) with TF-IDF vectorization.Brain: Google Gemini 1.5 Flash API for deep contextual analysis.

Challenges we ran into * Data Bias: Balancing the dataset to ensure the model didn't favor "Fake" or "Real" labels.API Security: Implementing st.secrets to keep our API keys hidden.Latency: Optimizing the hybrid workflow so results appear in seconds.

Accomplishments that we're proud of Successfully integrating a "Classic" ML model with a "Modern" LLM to create a Dual-Verification system that is more accurate than either tool alone.

What's next for TruthGuard Developing a browser extension to provide real-time "Trust Scores" on social media feeds and expanding the dataset to include multi-language support.

Built With

• python
• streamlit