TruthLens — ML-Powered Misinformation Detector
AI system that detects fake news, verifies facts, and explains decisions in real-time
Overview
TruthLens identifies misinformation in short-form news by combining a lightweight ML classifier with automatic fact-checking. It delivers real-time predictions, transparent explanations, confidence scores, and verified references — all inside a streamlined Streamlit dashboard.
Key Objectives
Build Reliable Detection
Lightweight ML classifier optimized for short-form news text.
Integrate Fact-Checking
Google Fact Check API with automatic fallback to cached local index.
Explain Decisions
Confidence scores and feature contributions build user trust.
Streamlit Dashboard
Multi-tab UI for predictions, facts, feedback, and performance metrics.
High Accuracy
~91.5% validation accuracy across REAL and FAKE classes.
Future-Ready
Modular architecture for APIs, embeddings, and active-learning loops.
Core Features
ML Classifier
TF-IDF n-grams + numeric features + Logistic Regression.
Fact-Checking Layer
Google Fact Check API with offline fallback cache.
Explainable AI
Shows feature influence + confidence for each prediction.
Performance Dashboard
Confusion matrix, precision, recall, and F1-score.
User Feedback
Reports questionable results for human review.
Modular Architecture
Easily integrates new APIs, embeddings, and pipelines.
Tech Stack
Planned Enhancements
More Fact-Check Sources
Extend beyond Google for broader coverage.
Transformer Embeddings
Improve text representation with modern NLP models.
Active-Learning
Continuously improve accuracy using user feedback.
Output Screenshots
