Semantic Catalog

Overview

The Semantic Catalog is the central place where Jedify interprets, organizes, and displays the customer's data warehouse schema as part of the semantic modeling workflow. It helps data experts understand how tables, columns, labels, and semantic entities connect to one another, and provides the foundation for grounding, semantic reasoning, and question interpretation.

The catalog automatically extracts structural metadata from the connected data warehouse, enriches it with semantic signals (labels, entity matches, grounding potential), and exposes it in a clean, navigable interface.

Audience

This page is designed for:

Data experts – maintain and grow the semantic model, add tables, validate metadata.
Administrators – govern catalog structure and approve changes.
Business users – explore semantics, understand entities, and validate interpretation paths.

Benefits

Provides a unified inventory of all schemas, tables, and columns Jedify has interpreted.
Surfaces automatic semantic signals, including PII, financial tags, and linked entities.
Helps identify structural drift using Check Changes.
Enables fast onboarding of new assets through Add Tables.
Enhances question-answering accuracy through groundable columns.
Offers easy navigation and clear previews to support semantic modeling decisions.

Key Concepts

Schemas, Tables, and Columns

The catalog mirrors your warehouse structure. Each level can be explored directly through the left-hand navigation pane.

Labels

Jedify automatically evaluates columns and tags them with labels such as pii or financial. These labels help guide compliance decisions and semantic prioritization.

Related Entities

Tables and columns may be linked to entities from the Semantic Fusion Model. These links help the system understand which semantic concepts your data represents. You can read more in the Semantic Fusion Model page.

Sample Data

Both tables and columns may show sample values, providing a quick sense of the kind of data stored. This is intentionally a small preview, not a full dataset.

Grounding

Grounding indicates that a column's values are categorial and query-mappable—meaning that phrases users write in natural-language questions can be reliably mapped back to values found in that column (e.g., "AMZ" → "Amazon").

Only columns can be enabled for grounding.

Filters

The top of the catalog includes filters:

Type: All, Schema, Table, Column
Grounding: All, Enabled, Not Enabled

These help narrow down exploration based on modeling needs.

Using the Semantic Catalog

Navigating the Catalog

The catalog is organized in a left-hand hierarchy:

Schema → Tables → Columns

Selecting a table or column reveals a detailed drawer on the right.

Table Drawer

Each table exposes:

Name
Description
Labels (e.g., financial, pii)
Related entities from the Semantic Fusion Model
Sample data preview (tabular snippet)

This view helps analysts quickly understand what the table represents and how it connects to semantic entities across the system.

Column Drawer

Each column exposes:

Name
Description
Type (e.g., STRING)
Labels
Related entities
Data sample (list of representative values)
Grounding toggle (enable/disable)

Columns carry the richest semantic signals and are central to downstream modeling and grounding decisions.

Permissions

Business users can explore the catalog
Data experts and administrators can:
Run Check Changes
Add tables
Manage semantic modeling assets

Best Practices

Use Check Changes regularly to ensure the catalog stays aligned with the warehouse.
Review labels and related entities as part of your semantic modeling workflow.
Enable grounding only when column values represent clear, categorial concepts.
Use table and column previews to validate interpretation before building Semantic Statements.