Taxonomy Reference
Taxonomy Reference
Section titled “Taxonomy Reference”What is the taxonomy?
Section titled “What is the taxonomy?”The taxonomy organises indicators into domains — coherent areas of organisational practice. It is stored as structured CSV tables in the NEI repository, not embedded in indicator definitions.
This separation means:
- Indicators can belong to multiple domains without changing their definitions
- Domain reorganizations do not require modifying indicator files
- Taxonomy can evolve on its own versioning cycle
Taxonomy structure
Section titled “Taxonomy structure”The taxonomy is stored as two CSV files per version:
Nodes table (NDT-<version>-nodes.csv)
Section titled “Nodes table (NDT-<version>-nodes.csv)”| Column | Description |
|---|---|
taxonomy_id | The taxonomy version this node belongs to |
node_id | Unique node identifier (e.g., NDT-p3wupm, NDI-ypwtie) |
node_type | root, domain, or indicator |
label | Human-readable name |
Edges table (NDT-<version>-edges.csv)
Section titled “Edges table (NDT-<version>-edges.csv)”| Column | Description |
|---|---|
taxonomy_id | The taxonomy version this edge belongs to |
source_id | Source node ID |
relationship | Relationship type |
target_id | Target node ID |
Relationship types
Section titled “Relationship types”| Relationship | Meaning |
|---|---|
parent_of | Root → domain (hierarchical containment) |
contains | Domain → indicator (membership; multi-domain allowed) |
related_to | Indicators addressing overlapping concerns |
mitigates | Structural indicator reduces an experiential signal |
reinforces | One indicator strengthens the effect of another |
depends_on | One indicator logically presupposes another |
Taxonomy versions
Section titled “Taxonomy versions”| Version | Status | Domains | Indicators |
|---|---|---|---|
| NDT-1.0.0 | Standard | 6 | 13 (as of March 2026) |
| NDT-0.1.0C-NDP-seed | Candidate | 15 | 13 active (11 retired March 2026) |
Identifier generation
Section titled “Identifier generation”Domain node IDs are generated deterministically from normalized domain names using the same algorithm as indicator IDs:
normalize_name → sha256 → base32 → lowercase → first 6 chars → prefix NDT-This means the same domain name always produces the same ID, making taxonomy merges and comparisons reliable.