Tool · Business Glossary

Business Glossary.

Where AI-suggested matches between business terms and database columns get reviewed, approved, and signed off.

216 business terms. 67 matched. 23 pending your review.

Semantic MatchingHuman ApprovalConflict DetectionMatch Scoring

Book a demo

216LDEs

67matched

23pending

Business Glossary

Pending review

23 pending

Customer ID

lde-001

Unique identifier assigned to each customer in the system

CRM Systemcustomers3 matches

Physical Matches:

CUST_ID(94%)customer_identifier(87%)cid(72%)

Physical Implementations

CUST_IDPerfect Match

94%

CRM Systemcustomers

customer_identifiersemantic

87%

CRM Systemcustomers

cidfuzzy

72%

CRM Systemcustomers

Account Balance

lde-002

Current monetary balance of the account

ERP Platformaccounts2 matches

Physical Matches:

ACCT_BAL(91%)balance_amount(85%)

Transaction Date

lde-003

Date when the transaction was processed

Payment Gatewaytransactions2 matches

Physical Matches:

TXN_DATE(96%)trans_dt(78%)

Human in the Loop

The Process

How semantic matching works.

Extract

Logical and Physical Data Elements are extracted from documents into structured inventories.

Embed

A sentence-transformer model generates semantic embeddings for every term.

Match

IDA finds the top three candidate matches per LDE, each with a confidence score.

Approve

Your team reviews and approves or rejects each suggestion — every decision is logged.

Match Quality

How confidence is scored.

Every match comes with a confidence score and a tier. The tier tells your reviewer how much attention each suggestion needs.

Two LDEs trying to claim the same PDE? IDA flags the conflict before you commit the match — you decide which is canonical.

Perfect Match

95–100%

Definition contains the exact business term. Auto-approveable with confidence.

Semantic Match

85–94%

High semantic similarity between business term and column meaning.

Fuzzy Match

70–84%

Partial similarity. Always needs human review.

Review Tools

Everything a glossary reviewer needs.

Search & filter

Find terms by name, definition, system, or match status.

Group by anything

Organise by system, table, or flat list — your call.

Bulk actions

Approve or reject hundreds of matches in one pass.

Conflict alerts

Automatic detection when two LDEs map to the same PDE.

Audit trail

Full history of who approved what, and when.

Re-process

Re-run matching as new documents land. The glossary stays current.

Built From

What feeds the glossary.

Every match has two sides. The glossary pulls them from these two inventories.

Logical Data Elements

The business terms being matched.

Explore inventory

Physical Data Elements

The database columns being matched to.

Explore inventory

Companion Tools

What lives downstream.

Approved matches in the Business Glossary feed both the enterprise-wide Global Glossary and the Data Lineage graph.

Global Glossary

Enterprise-wide consolidation of approved terms.

Explore

Data Lineage

Interactive flow graph from source to consumer.

Explore

Get Started

Ready to build your verified glossary?

We'll run IDA against your dictionaries and show you the approve-queue in a 30-minute call.

Book a demo