Upload 10 files

datasets/atoms&edges.graphs/atoms&edges.graphsREADME.md

@@ -0,0 +1,2 @@
+# Datasets (Atoms)
+This folder contains the **atomized datasets** used to build the Master Grid.Each `atoms*_v2.csv` file is a **node table** (atoms) extracted from a source CSV.## Common atom schema (all atoms*.csv)Each row is an atom (node) with:- `atom_id` — stable unique ID- `atom_category` — `rule | pattern | value | reference`- `atom_type` — subtype (e.g. `domain_profile`, `template_profile`)- `primary_value` — main human-readable value- `attributes_json` — original fields + provenance- `tags_json` — retrieval-friendly tags- `source_dataset_id`, `source_row_id`, `source_line_no` — lineage---## atoms1_v2.csv — Tone Policy Registry**Purpose:** tone/voice policies for prompt compilation.Includes:- tone profiles, voice keywords- do/don’t directives- formatting cues- certainty calibration rules## atoms2_v2.csv — Domain Policy Registry**Purpose:** domain-level defaults and required sections.Includes:- domain profiles- required sections per domain- default template reference per domain- risk baseline per domain## atoms3_v2.csv — Expansion Policy Registry**Purpose:** how to handle missing inputs (fallbacks/defaults).Includes:- expand policy profiles- missing input types- default actions + default values- extracted tags for routing## atoms4_v2.csv — Prompt Template Catalogue**Purpose:** concrete templates per domain with constraints + output format.Includes:- template profiles- required sections- directives + constraints- output format cues- tone policy references- risk baseline## atoms5_v2.csv — Command / Mode Registry**Purpose:** runtime command catalog for the compiler (modes, phases, outputs).Includes:- command profiles- aliases- RAG keywords- safety notes- phase/category/output_type atoms## atoms6_v2.csv — Latent Constraints Signals**Purpose:** trigger→action signals (inject constraints when certain cues appear).Includes:- signal profiles- trigger hints- recommended actions- severity + domain/type labels## atoms7_v2.csv — Output Patterns Registry**Purpose:** reusable output “skeletons” / formatting patterns by output type.Includes:- output pattern profiles- skeleton (structure)- notes (usage guidance)- risk baseline- output type## atoms8_v2.csv — Validation Rules Registry**Purpose:** enforceable tone/format validation policies (validator view).Includes:- tone profiles (validator)- do/don’t directives- formatting cues- certainty calibration rules- verbosity + risk baseline---## How these datasets are used- Treat atoms as **retrievable nodes** (vector DB / search index).- Use `grid/edges_v1.csv` to **expand context** across datasets (graph hops).- For a single-table node load, use `grid/mastergrid_nodes_v1.csv`.