Upload viral_muse_dataset_card.md

datasets./viral_muse_dataset_card.md

@@ -0,0 +1,107 @@
+---
+dataset_name: viral-muse-vectorized-kg
+pretty_name: Viral Muse Vectorized Knowledge Graph
+version: v1.0.0
+created_at_utc: 2025-12-13
+license: CC-BY-4.0
+tags:
+  - knowledge-graph
+  - master-grid
+  - synthetic-data
+  - retrieval
+  - agentic
+  - content-patterns
+task_categories:
+  - text-retrieval
+  - text-generation
+languages:
+  - en
+---
+
+# Viral Muse Vectorized Knowledge Graph (KG)
+
+This dataset is a **synthetic, structured Master Grid** built for retrieval + linking workflows.
+It contains:
+
+- **Atoms** (nodes): normalized semantic units derived from source rows
+- **Edges** (relations): deterministic internal links, platform-alignment links, and rule-based semantic links
+- **Knowledge Map**: dataset registry (schemas, atomization strategy, enabled rules)
+
+## Files
+
+- `viral_muse_atoms_master_v1.csv`
+- `viral_muse_edges_master_v1.csv`
+- `viral_muse_knowledge_map_v1.csv`
+- `viral_muse_kg_manifest_v1.json`
+
+## What’s inside
+
+### Atom counts
+
+{
+  "creative_partner_advice_map": 406,
+  "genre_transformation_rules": 169,
+  "lyric_structure_map": 300,
+  "tiktok_concept_patterns": 299,
+  "viral_pattern_signals": 308,
+  "viral_potential_rated": 616
+}
+
+Total atoms: **2098**
+
+### Edge counts by relation type
+
+{
+  "guided_by": 203,
+  "has_bundle": 308,
+  "platform_aligned": 7440,
+  "platform_multi_aligned": 260,
+  "semantic_related": 857
+}
+
+Total edges: **9068**
+
+## Schema
+
+### atoms_master columns
+
+atom_id, source_dataset_id, source_row_id, atom_category, primary_label, secondary_labels, attributes, tags, provenance, confidence, status, notes
+
+### edges_master columns
+
+edge_id, from_atom_id, to_atom_id, relation_type, relation_subtype, weight, confidence, source_rule_ids, notes
+
+## Intended use
+
+Use this KG to:
+
+- retrieve “best matching” patterns (atoms) for a given request
+- follow relationships to adjacent patterns (edges) for expansion
+- power ranking, recommendation, and structured reasoning inside agent pipelines
+
+## Recommended ingestion pattern
+
+1) **Vector DB**: upsert **atoms** as documents
+- Vector ID recommendation: `viral_muse:<source_dataset_id>:<atom_id>`
+- Attach metadata: `source_dataset_id`, `source_row_id`, `atom_category`, `confidence`, `status`, `version`
+
+2) **Graph layer**: ingest **edges** into a graph store (or adjacency index)
+- Use `edge_id` as primary key
+- Validate that `from_atom_id` and `to_atom_id` exist in atoms
+
+## Data origin & safety
+
+- **Synthetic**: not collected from real users
+- **No personal data**
+- **Non-operational**: patterns are for creative/retrieval workflows
+
+## Limitations
+
+- CPA semantic connectivity is intentionally conservative (low shared vocabulary)
+- Genre transformation rules (GTR) are present as atoms but have minimal linking in v1 (to be expanded with transformation-specific rules)
+
+## Citation / attribution
+
+If you use this dataset publicly, attribute:
+**Agentarium / Viral Muse (synthetic)** (CC-BY-4.0)
+