NollyP/nollypulse-schema-docs

Building the canonical intelligence and structured data infrastructure for African screen content NollyPulse Data Infrastructure & Canonical Schema Welcome to the official technical documentation and open schema repository for NollyPulse Technologies Limited. NollyPulse is building the canonical intelligence layer and structured data infrastructure for African screen content (Nollywood, regional TV, and independent African film ecosystems). 🗺️ System Overview The African entertainment industry is one of the largest under-documented creative ecosystems in the world. Key structural challenges include: - Unstructured, fragmented content metadata across disparate web sources. - Ambiguous entity relationships (cast aliases, localized title variants across African languages like Yoruba, Swahili, Zulu, Hausa, Igbo, Amharic, etc.) - Absence of standardized canonical content identifiers. - High rates of hallucination in Large Language Models (LLMs) when querying African media history. NollyPulse addresses this discovery gap by providing a unified, multi-stage ingestion, entity-resolution, and graph-relational pipeline designed to serve RAG-ready, verified datasets to streaming platforms, researchers, and global AI systems. 🏗️ Architecture & Data Pipeline [ Raw Web Sources / NFVCB Registries ] │ ▼ [ Automated Ingestion Pipeline (Python) ] │ ▼ [ NLP Entity Resolution & Deduplication ] │ ▼ [ Canonical Graph & Relational Schema (PostgreSQL) ] │ ▼ [ REST / GraphQL API Layer & RAG Datasets ] Key Engineering Pillars 1. Automated Ingestion: Custom scraping and ingestion services aggregating raw titles, credits, and historical archives dating back to the 1960s, alongside official censorship registry logs dating back to 2017. 2. Entity Resolution: Heuristic and NLP matching algorithms designed to deduplicate records, normalize regional language titles, and map actor/director aliases. 3. Canonical Identifiers: Unique, persistent ID assignment (`NP-TITLE-XXXXX`, `NP-PERSON-XXXXX`) establishing ground-truth relationships across titles, crew, and distribution entities. 4. AI/L …