📖 About the Project
🔍 Inspiration
In modern politics, promises are often forgotten or distorted once elections are over. I wanted to build a transparent system where anyone can verify political pledges and track their fulfillment over time.
The inspiration came from seeing fact-checking organizations struggle with scalability — and realizing AI could automate much of that process, while blockchain could guarantee trust and immutability.
📚 What I Learned
Through this project, I learned:
- How Natural Language Processing (NLP) models can match campaign promises with real-world actions.
- How news APIs and government open data can be integrated into an automated tracking pipeline.
- How blockchain’s immutability and decentralization can be leveraged to build tamper-proof public records.
- The importance of designing for user trust and clear visualization in civic tech.
🛠️ How I Built It
- Data Collection
- Gathered political pledge data from open election platforms.
- Used news APIs (e.g., GDELT, NewsAPI) to fetch relevant articles and announcements.
- Gathered political pledge data from open election platforms.
- AI Analysis
- Employed an LLM-based classifier to map news/events to specific pledges.
- Calculated a fulfillment score for each promise.
- Employed an LLM-based classifier to map news/events to specific pledges.
- Blockchain Recording
- Recorded fulfillment data on a public blockchain (Polygon testnet in the prototype).
- Used smart contracts to store pledge ID, timestamp, and fulfillment status.
- Recorded fulfillment data on a public blockchain (Polygon testnet in the prototype).
- Frontend Visualization
- Built a simple web dashboard using HTML, CSS, and JavaScript to display:
- Candidate list
- Promises and their fulfillment scores
- Immutable blockchain transaction IDs for verification
- Candidate list
- Built a simple web dashboard using HTML, CSS, and JavaScript to display:
🚧 Challenges
- Data Quality
Political statements often lack standardization, making NLP matching non-trivial. - Entity Disambiguation
Ensuring that a news article’s “policy” matches the exact pledge and candidate required careful prompt engineering and model tuning. - Blockchain Cost & Speed
Deciding how frequently to write to the blockchain without incurring excessive gas fees. - User Experience
Balancing transparency (showing blockchain hashes) with accessibility for non-technical users.
✨ Next Steps
- Integrate multilingual support for global use.
- Add community fact-checking with governance tokens to allow users to challenge or confirm AI results.
- Expand to include economic and environmental policy tracking.
Log in or sign up for Devpost to join the conversation.