Alexandria Aeternum: A Provenance-Verified Semantic Dataset for AI Training

Alexandria Aeternum is a curated corpus of 10,097 museum-verified artworks from the Metropolitan Museum of Art Open Access Initiative, enriched with AI-generated semantic metadata following the Golden Codex standard. Contents Full dataset with images: The complete dataset including high-resolution images is available on HuggingFace: huggingface.co/datasets/Metavolve-Labs/alexandria-aeternum-10k alexandria_aeternum_10k.parquet (14MB) - Structured metadata only for ML pipelines Each artwork includes 50+ metadata fields: Visual Analysis: Composition, color harmony, technique mastery, light/shadow, spatial depth Emotional Mapping: Primary/secondary emotions, viewer transformation, mood, narrative arc Poetic Interpretation: Mythological resonance, eternal voice, symbolic depth Cultural Context: Historical significance, style, period, medium Provenance Verification: SHA-256 soulmark hash, C2PA compatibility Embedded Metadata (PEST Framework) Every image contains its own soul. The full Golden Codex metadata is embedded directly into each JPG file using XMP (ISO 16684-1). This means the provenance and semantic context travel WITH the image—surviving downloads, shares, and re-uploads. To extract the embedded metadata from any image: pip install golden-codex-reader from golden_codex_reader import extract_codex metadata = extract_codex("image.jpg") print(metadata["poetic_interpretation"]["eternal_voice"]) See the Golden Codex Reader SDK for full documentation. Purpose Designed for AI training with built-in provenance verification, this dataset demonstrates the Human-AI Hybrid Curation paradigm: human-originated artifacts (preserving entropy) enriched with AI-generated semantic scaffolding (providing density) to create Super-Ground Truth training data. Related Resources Hugging Face Dataset Interactive Explorer Verification SDK