TrustLens is an AI-powered web application that helps users verify the authenticity of news articles, detect political and emotional bias, and compare the same story across multiple sources β all in one place.
βDonβt just read the news. Understand it.β
- Classifies articles as Real / Misleading / Fake
- Provides a confidence score
- Highlights suspicious phrases
- Political bias: Left / Right / Neutral
- Emotional bias: Fear, anger, joy, sadness, neutral
- Linguistic bias: Clickbait, exaggeration, loaded words
Finds the same story from multiple publishers Compares:
- Headlines
- Tone
- Claims
- Missing information
- Shows how narratives differ
** A final score (0β100) based on: ** Accuracy Bias neutrality Source credibility Emotional manipulation
Instead of just showing numbers, TrustLens explains:
Why this article may be misleading ? What language is manipulative ? What other sources say ?
React / Next.js Tailwind CSS Chart.js / Recharts
Python FastAPI / Flask
HuggingFace Transformers spaCy TextBlob Scikit-learn
GNews API (Free tier) NewsAPI (Free tier) Web scraping (fallback)
Frontend: Vercel / Netlify Backend: Render / Railway
π§© System Architecture Frontend (React) β Backend API (FastAPI) β AI Models (NLP) β News APIs + Scrapers
π§ͺ How It Works
User pastes a news article link or text Backend extracts and cleans the content
Authenticity Political bias Emotional tone
System fetches similar news from other sources
Articles are compared A final trust score is calculated Results are displayed with explanations
π₯ Installation (Local Setup)
-
Clone the repository
git clone https://github.com/your-username/trustlens.git
cd trustlens -
Frontend Setup
cd frontend
npm install
npm run dev -
Backend Setup
cd backend
python -m venv env
source env/bin/activate # On Windows: env\Scripts\activate
pip install -r requirements.txt
python app.py
Misinformation affects:
- Elections
- Public opinion
- Mental health
- Social harmony
TrustLens promotes: β Transparency β Critical thinking β Media literacy β Responsible AI
TrustLens does not censor content. It provides analysis, not authority. Final judgment is always left to the user.