Oracle & Verification Layer

1 Multi-Source Oracle Architecture

The Merin oracle design adopts a multi-source + AI arbitration + ZK verification structure to maximize resistance to manipulation and ensure data accuracy.

Core components include:

• Primary Oracles: Standardized data sources such as Chainlink and UMA;

• AI-Signal Oracles: Merin’s proprietary AI agents that parse natural language news, social media, and on-chain data;

• Community Validators: DAO-level user voting confirmation;

• ZK-Proof Verifier: Ensures data input consistency via zero-knowledge proofs.

Data flow process:

External Data Sources → AI Signal Processing → Oracle Aggregation → zk Verification → Settlement Layer

---

2 AI-Augmented Fact Validation

Merin introduces NLP + Embedding Retrieval Models to automatically parse natural language events such as:

“Will BTC reach $100,000 by June 2026?”

AI agents crawl reliable sources (news outlets, exchange data, regulatory announcements, etc.) to generate structured judgments with confidence scores. These scores are then combined with on-chain voting outcomes to form a weighted consensus determination mechanism.

---

3 Dispute Resolution and Economic Security

When oracle conflicts or disputes occur, the system enters a secondary arbitration phase:

• Activation of the Merin Resolution DAO;

• $MRN token holders can stake to participate in adjudication;

• The economic slashing mechanism ensures correct voting incentives;

• The entire process is protected by the zk-Vote module, ensuring voting privacy and result verifiability.