The Glossary Builder - Lingo - Governance Accelerator

Organizations frequently reuse the same business metrics across dashboards, reports, and analytical models. However, these metrics are often defined differently in different places, creating reporting conflicts and reducing trust in analytics. This recipe analyzes reporting columns, traces lineage to upstream sources, identifies business metric candidates, detects formula conflicts, and automatically prepares governance-ready glossary terms with suggested domains and descriptions.

What you receive

How the recipe works (guided flow)

Step 1 — Reporting Surface
Joins the chart and chartchild tables to build a base inventory of every report column across all dashboards.

Step 2 — Lineage Presence
Flags each report column with whether it has any resolvable upstream lineage, without yet resolving the source.

Step 3 — Term Normalization
Lowercases and collapses separators in each column name into a structural "normalized_term" grouping key.

Step 4 — Usage Impact
Aggregates how many report columns share each normalized term to measure its usage footprint.

Step 5 — Term Candidates
Filters out technical identifiers, keys, codes, and date/time fields to surface genuine business term candidates.

Step 6 — Identifying Criticality
Combines lineage presence, business-term candidacy, and usage count into a single CRITICAL_FLAG per column.

Step 7 — Critical Scope
Narrows the working set down to only the columns flagged as governance-critical.

Step 8 — Immediate Sources
Resolves the single deterministic upstream table and column feeding each critical report column.

Step 9 — Source Signature
Builds a standardized "table.column" source signature for each critical column, marking unresolved sources explicitly.

Step 10 — Formula Conflicts
Normalizes formulas and counts distinct formula variants per term to detect true definition conflicts.

Step 11 — Conflict Details
Pulls every report column contributing to a detected formula conflict, alongside its formula and source signature.

Step 12 — Conflict Details and Governance Roles Associated
Enriches each conflicting column with its owning report's governance roles (owner, steward, custodian) for review.

Step 13 — Summary of Conflicts
Presents the full formula-conflict dataset, joined with governance context, as a reviewable summary.

Step 14 — Non-Conflicts
Isolates terms with a single consistent formula and no detected conflict across all reports.

Step 15 — Unique Definitions
Deduplicates non-conflicting terms down to one clean formula per normalized term.

Step 16 — Locating Existing Domains
Pulls the set of existing Business Glossary domains eligible for term classification from the domain catalog.

Step 17 — Domain Context
Collapses all available glossary domains into a single reference row for downstream AI classification.

Step 18 — Domain Mapping
Attaches the full list of available domains to every normalized term and formula pending classification.

Step 19 — AI Suggest Logical Domain
AI reviews each term, formula, and the domain list to recommend the best-fit business glossary domain.

Step 20 — Displaying AI Suggestion
Surfaces the AI's suggested domain per term for review, along with a row-and-column summary of the results.

Step 21 — Fetching Domain ID
Matches each AI-suggested domain name back to its glossary domain ID, flagging any suggestions that don't resolve.

Step 22 — AI Suggest Description Using Formula and Term Name
AI drafts a glossary-ready business description for each term based on its name and underlying formula.

Step 23 — Term Creation
Compiles the final term, domain ID, and description set for governance teams to review and publish as glossary entries, flagging any rows still missing a resolved domain.

Sample insights (illustrative)

Make sure the following ingredients are available in your workspace:

• Connected datasets are crawled and profiled
• Columns have distinct counts and top values generated