# EnterpriseGovernance — Schema

> Synthetic enterprise data-governance snapshot across 12 business units. Point-in-time with 24-month operational-signal trajectory and annual self-assessment.

## Tables

### `business_units.csv` (12 rows)

| Column | Type | Notes |
| --- | --- | --- |
| `bu_id` | string | Primary key. |
| `bu_name` | string | Display name. |
| `business_criticality` | string | `critical`, `high`, `medium`, `low`. |
| `has_dedicated_data_steward` | boolean | **Bias signal**: 4 of 12 BUs have no steward (IT director substitutes). |
| `steward_role_if_substitute` | string | Empty if BU has dedicated steward; otherwise role of substitute. |
| `employee_count` | integer | Drives asset count. |

### `capabilities.csv` (8 rows)

| Column | Type | Notes |
| --- | --- | --- |
| `capability_id` | string | Primary key. |
| `capability_name` | string | One of: Catalog, Lineage, Quality, Access, Retention, Classification, Stewardship, Data Literacy. |
| `weight_in_aggregate_score` | decimal | Used to compute weighted enterprise capability score. |
| `target_level` | integer (1-5) | Maturity-model target for each capability. |

### `assets.csv` (~6,300 rows)

| Column | Type | Notes |
| --- | --- | --- |
| `asset_id` | string | Primary key. Format `AST-NNNNN`. |
| `bu_id` | string | FK → `business_units.bu_id`. |
| `asset_type` | string | One of 7 types (Semantic Model, Report, Dashboard, etc.). |
| `certification_status` | string | `Certified`, `Promoted`, `Master Data`, `Untagged`. Distribution strongly biased by `has_dedicated_data_steward`. |
| `in_enterprise_catalog` | boolean | ~78% true for stewarded BUs, ~42% for non-stewarded. |
| `has_lineage_documented` | boolean | Subset of catalog presence. |
| `has_data_classification` | boolean | Independent of catalog. |
| `last_review_date` | date | When the asset was last reviewed. |

### `capability_scores.csv` (~2,300 rows)

The **operational-signal** view of capability — the unbiased input.

| Column | Type | Notes |
| --- | --- | --- |
| `month` | date (ISO) | First of month, 24 months ending 2026-05. |
| `bu_id` | string | FK. |
| `capability_id` | string | FK. |
| `capability_level` | decimal (1.0-5.0) | Maturity score derived from operational signals. |
| `source` | string | Constant `operational-signal` for this table. |

### `self_assessment.csv` (96 rows)

The **biased input** the dashboard exists to correct.

| Column | Type | Notes |
| --- | --- | --- |
| `assessment_date` | date | Most recent annual survey: `2026-05-01`. |
| `bu_id` | string | FK. |
| `capability_id` | string | FK. |
| `self_rated_level` | decimal (1.0-5.0) | **Systematically biased**: stewarded BUs over-rate by +0.3 to +1.0; non-stewarded BUs are honest or slightly under-rate. |
| `source` | string | Constant `self-assessment-survey`. |
| `respondent_role` | string | `data-steward` or `IT-director-substitute`. |

## The dataset's reason for existing

The bias in `self_assessment.csv` versus the truth in `capability_scores.csv` is the entire reason this dataset is interesting. The triangulation — operational signal + self-assessment + the gap between them — is exactly what the [EnterpriseGovernance starter BRD](../../appendix-e/enterprisegovernance-starter.md) is about.

A chapter exercise that uses self-assessment as ground truth is teaching the wrong discipline.

## Suggested chapter exercises

- **Chapter 11 (prep for AI)**: Author the AI instructions that tell Copilot to weight operational signals higher than self-assessment when ranking gap severity.
- **Chapter 13 (when AI misleads)**: Ask Copilot "which BU has the largest governance gap?" without instructions. Watch whether it uses self-assessment (where the answer favors non-stewarded BUs because they self-rate low) or triangulates (where the answer reveals the stewarded BUs over-claiming).
- **Chapter 14 (responsible AI)**: This dataset embeds the canonical responsible-AI question — *whose view of reality drives the recommendation?* Use it to teach explicit triangulation as an ethical discipline, not a technical one.

## Reproducibility

Generated by `scripts/datasets/generate-enterprisegovernance.js`. Source seed: `0xE17EC0B0`.