Skip to content

πŸš€ On-Chain Election Promise Verification Platform

On-Chain Election Promise Verification Platform
(AI-powered system to track election promise fulfillment and record results on blockchain)

πŸ—‚οΈ Project Overview

  • Student ID: 20210403
  • Name: Jihyun Yoo
  • Project Title: On-Chain Election Promise Verification Platform
  • Summary (3–4 sentences):
    This platform transforms election promises into structured claims, collects evidence from news articles, press releases, and public datasets, and uses AI to automatically match and evaluate status and progress.
    The core results (status, progress, confidence, version hash) and evidence bundle hash (IPFS/Arweave) are stored on the blockchain (EVM) to ensure immutability and auditability.
    Anyone can view a dashboard showing candidate, party, or policy-specific progress, and disputes are handled transparently through DAO governance (Snapshot).
    The goal is to create a fact-based, auditable record of election promise fulfillment, free from political bias.

1. 🧩 Problem: What Problem Are You Solving?

image

  • Real-world issue:
    After elections, the progress of campaign promises is scattered across fragmented news and government announcements, with a high risk of bias or misrepresentation.
    Voters lack an easy, consolidated view, while journalists and NGOs struggle to maintain consistent, evidence-linked evaluations.
  • Limitations of existing solutions:
  • PR-focused performance reports often lack traceability
  • Spreadsheet or document-based tracking lacks version control and auditability
  • Existing fact-checking platforms rarely integrate on-chain immutability and governance-based corrections
  • Why it matters:
    Standardizing promise tracking with evidence-based, tamper-proof records allows the public, media, and researchers to debate from the same dataset, enhancing democratic accountability.

2. πŸ’‘ Solution: Your Proposed Approach

  • Core idea:
    1) Extract structured claims (action, target, KPI, deadline) from campaign promises using Claim Extraction.
    2) Collect and normalize evidence from news, press releases, and government datasets, scoring trustworthiness (C2PA signatures, publisher credibility, source diversity).
    3) Calculate progress (based on KPI achievement rates) and confidence (based on source credibility, consistency, freshness).
    4) Store the evidence bundle on IPFS/Arweave, and commit the root hash (Merkle/DAG) and summary (status, progress, confidence, model version) on-chain.
    5) Handle disputes via DAO governance (Snapshot) with transparent challenge and update flows.
  • AI & Blockchain integration:
  • AI: NLP-based claim extraction, entity linking, event detection, evidence scoring with hybrid rule+LLM methods
  • Blockchain: Immutable storage of result summaries and evidence hashes, governance records for dispute resolution
  • Architecture Diagram (ASCII):

3. πŸ”— Why Blockchain (and Token)?

  • Why blockchain is needed:
  • Immutability: Locks evaluation summaries, versions, and evidence hashes in an unalterable record
  • Auditability: Records who/when/what was uploaded with on-chain timestamps
  • Governance: Uses on-chain voting (Snapshot) for disputes and corrections
  • Problem blockchain solves:
    Removes reliance on a single trusted operator and guarantees permanent traceability of evaluation history.
  • Token design (optional):
  • Initial phase: No token, governance via Snapshot + Reputation (Soulbound) system
  • Extended phase: Introduce small stake & slash mechanism for fact-checkers, and governance/reputation token if needed
  • Why optional: For early-stage public/academic prototypes, blockchain can serve its purpose without a native token

4. πŸ› οΈ MVP or Prototype

Current status:
β˜‘ Prototype

Key Features

  • Scope: 1 pilot region, 3 policy areas, 2–3 candidates
  • Data: Gov portals + press releases + news RSS (Python + Prefect)
  • AI: NLP for promise–event matching & scoring (trust, freshness, consistency)
  • Blockchain: Evidence β†’ IPFS, summary/rootHash β†’ Base/Kaia testnet
  • UI: Filters, status cards, evidence links, on-chain TX

Stack

Python Β· Prefect Β· HuggingFace Β· PostgreSQL Β· IPFS Β· Next.js Β· Tailwind Β· Snapshot

Repo: https://codesandbox.io/p/sandbox/r5fdxt
Screens: Promise list Β· Detail view Β· Voting screen

5. πŸ“¬ Submission to Hackathons or Grant Programs

  • Name of program submitted to: QuSat Group Hackathon 2025
  • Submission link: (URL)
  • Short summary:
    β€œAn AI-powered platform that maps campaign promises to evidence, calculates KPI-based progress, and records results immutably on-chain, with DAO governance for dispute resolution.”
  • (Optional) Screenshot: image

6. πŸ€” Reflection & Future Work

  • What I learned:
  • gas fee matters!
  • Political/policy text often contains ambiguity and this is very high-stakes domain, to make it reliable -> deliberate design choices

Future Work - If possible, Implement with Korean sample political sources and publish public.

7. πŸ“š References

ν•œκ΅­λ§ˆλ‹ˆνŽ˜μŠ€ν† μ‹€μ²œλ³ΈλΆ€ http://manifesto.or.kr/?p=10407