Inspiration The inspiration for TruePulse came from the growing concern about misinformation and fake news spreading rapidly across social media and online platforms. In an era where anyone can publish content online, it became increasingly difficult for people to distinguish between credible journalism and fabricated stories. The project was inspired by the need to democratize fact-checking tools, making AI-powered news verification accessible to everyday users rather than just professional journalists and fact-checkers.

What it does TruePulse is an intelligent news verification platform that provides comprehensive analysis of online articles through:

AI-Powered Classification: Uses advanced Hugging Face transformer models (RoBERTa-based) to classify articles as "Real," "Fake," or "Uncertain" Sentiment Analysis: Analyzes the emotional tone of articles to detect potential bias or sensationalism Trust Score Calculation: Generates a 0-100 credibility score based on multiple factors including AI classification, source domain reputation, and sentiment analysis External Verification: Cross-references articles with trusted news sources using the News API to provide additional context Smart Content Extraction: Automatically extracts and processes article content from URLs using web scraping User-Friendly Interface: Presents complex AI analysis in an easily digestible format with detailed reasoning How we built it TruePulse was built using a modern full-stack architecture:

Backend (Python/Flask):

Flask API with modular route structure (/routes/analyze.py, /routes/health.py) Custom FakeNewsClassifier class leveraging Hugging Face transformers Primary model: winterForestStump/Roberta-fake-news-detector with fallback to akhooli/fake-news-detection-bert Sentiment analysis using cardiffnlp/twitter-roberta-base-sentiment-latest Web scraping with BeautifulSoup for content extraction News API integration for external verification Intelligent text preprocessing and tokenization Frontend (React/TypeScript):

Modern React application built with Vite for fast development TypeScript for type safety and better development experience Custom hooks (useNewsAnalysis.ts) for state management Comprehensive UI component library using shadcn/ui Tailwind CSS for responsive, modern styling Theme switching capabilities (dark/light mode) Infrastructure:

CORS-enabled API for seamless frontend-backend communication Environment variable configuration for API keys Automated startup script (start.sh) for concurrent development Vercel deployment configuration Challenges we ran into Model Loading and Performance: Initially faced challenges with model loading times and memory usage. Implemented fallback mechanisms and model caching to ensure reliability.

Text Processing Limitations: Transformer models have token limits (512 tokens). Solved this by implementing intelligent text truncation using tokenizers while preserving context.

False Positive Sentiment Detection: The sentiment analysis was incorrectly classifying tragic news (accidents, disasters) as positive. Implemented keyword-based override logic with a comprehensive list of tragic keywords.

API Rate Limiting: External news API calls needed to be managed efficiently. Implemented proper error handling and fallback mechanisms.

Cross-Platform Deployment: Ensuring the application works consistently across different environments required careful dependency management and environment configuration.

Accomplishments that we're proud of Robust AI Pipeline: Successfully integrated multiple state-of-the-art NLP models with intelligent fallback mechanisms, achieving reliable fake news detection.

Smart Context Awareness: Implemented sophisticated text analysis that goes beyond simple classification, including sentiment override for tragic news and confidence-based uncertainty detection.

User Experience: Created an intuitive interface that makes complex AI analysis accessible to non-technical users while providing detailed explanations.

Modular Architecture: Built a scalable, maintainable codebase with clear separation of concerns and comprehensive error handling.

Production-Ready Features: Implemented proper logging, health checks, CORS handling, and deployment configurations.

What we learned AI Model Integration: Gained deep understanding of transformer models, tokenization, and the importance of proper text preprocessing for NLP tasks.

Balancing Accuracy and User Experience: Learned that raw AI predictions need human-interpretable explanations and confidence measures.

Handling Model Limitations: Discovered the importance of fallback mechanisms and graceful degradation when dealing with external AI models.

Full-Stack Development: Enhanced skills in connecting React frontends with Flask backends, managing state across components, and handling asynchronous operations.

Content Analysis Challenges: Learned about the complexity of news verification, including the need to consider source credibility, sentiment bias, and external validation.

What's next for TruePulse Short-term improvements:

Enhanced domain reputation scoring using comprehensive news source databases Real-time fact-checking integration with established fact-checking organizations Browser extension for instant article verification while browsing Mobile application for on-the-go news verification Advanced features:

Historical tracking of news stories to identify evolving narratives Social media integration to analyze viral content and misinformation spread Custom model training on domain-specific datasets for improved accuracy Multi-language support for global news verification Community features:

User feedback system to improve model accuracy through crowd-sourcing Journalist dashboard with advanced analytics and batch processing Educational resources about media literacy and critical thinking API access for researchers and developers to build upon the platform Technical enhancements:

Real-time processing with WebSocket connections Advanced caching mechanisms for improved performance Machine learning pipeline for continuous model improvement Integration with blockchain for immutable verification records TruePulse represents a significant step toward democratizing access to AI-powered fact-checking tools, and the roadmap focuses on making it an essential tool for combating misinformation in our digital age.

Built With

Share this project:

Updates

Submission history