📄 Unstructured Data Guide

High-fidelity extraction from non-tabular sources (PDF, Doc, Web) using LLM-gated scrapers and OCR pipelines. This sector enables the engine to process raw, natural language source material.

Status: HARDENED

• Architecture: LLM-Gated Scraper Fleet.
• Verification: 100% Parity.

Data Sources

• Web-based Acquisition: Universal scraper fleet (API-compatible).
• PDF/OCR: Textual extraction from institutional reports and documentation.

🛠️ Schema (StandardSchema)

• source_uri: URL or File path.
• content_type: text, image, or hybrid.
• metric: relevance_score, data_density, or sentiment.
• value: Numerical reading.
• integrity_hash: SHA-256 hash of the raw source.

---

Back to Index | User Manual | Data Veracity