Skip to content

[Optional] Creating an Ontology in Microsoft Fabric (Preview)

This guide walks you through creating an Ontology (preview) item in Microsoft Fabric, using the Build directly from OneLake approach. The ontology represents the Network Operations scenario (telecommunications), tracking outages and trouble tickets.

Schema Overview

The ontology is based on the schema defined in data/default/config/ontology_config.json:

Table Columns Key Description
tickets ticket_id, customer_name, issue_description, priority, status ticket_id Stores trouble ticket records
inspections inspection_id, ticket_id, result, score inspection_id Stores inspection results linked to tickets

Relationship: inspections → tickets (linked via ticket_id)

Prerequisites

  • A Microsoft Fabric workspace with Ontology (preview) enabled at the tenant level.
  • A Lakehouse in your workspace containing the tickets and inspections tables loaded into OneLake.
  • Appropriate permissions to create items in the workspace.

Step 1: Create the Ontology Item

  1. In your Fabric workspace, select + New item. Select New item

  2. Search for and select Ontology (preview).

Select ontology item 3. Enter a name for your ontology (e.g., NetworkOperationsOntology) & Click on Create.

Tip: Ontology names can include numbers, letters, and underscores. Do not use spaces or dashes.

Empty ontology canvas 4. The ontology opens when it's ready.

Step 2: Create Entity Types and Data Bindings

Entity types represent categories of objects in your business domain. For this schema, you will create two entity types: Tickets and Inspections.

2.1 — Add the Tickets Entity Type

  1. From the top ribbon or the center of the configuration canvas, select Add entity type.

Add entity type 2. Enter Tickets as the name and select Add Entity Type. Create entity type 3. The Tickets entity type appears on the configuration canvas and the Entity type configuration pane opens.

Tickets entity type configuration 4. Switch to the Bindings tab and select Add data to entity type.

Bindings tab 5. Choose your data source: a. Select your Lakehouse and select Connect. Select tickets data source b. Select the tickets table and select Next. Select tickets data source 6. Configure a Static data binding:

  • For Binding type, keep the default Static.

  • Under Bind your properties, the columns from the tickets table populate automatically:

    Source Column Property Name Type
    ticket_id ticket_id String
    customer_name customer_name String
    issue_description issue_description String
    priority priority String
    status status String

  • Select Save.

Tickets binding configuration 7. Back in the Entity type configuration pane, select Add entity type key. Tickets entity type key 8. Select ticket_id as the key property and select Save.

Tickets entity type key

2.2 — Add the Inspections Entity Type

Follow the same steps as above for the Inspections entity type:

  1. Select Add entity type from the ribbon. Add entity type 2
  2. Enter Inspections as the name and select Add Entity Type. Create entity type
  3. Switch to the Bindings tab → Add data to entity type. Bindings tab
  4. Choose your data source. a. Select your Lakehouse and select Connect. Select inspections data source b. Select the inspections table and select Next.

Select inspections table 5. Configure a Static data binding with the following columns:

Source Column Property Name Type
inspection_id inspection_id String
ticket_id ticket_id String
result result String
score score BigInt
  • Select Save.

Inspections binding configuration 6. Select Add entity type key → choose inspection_id.

Inspections entity type key 7. Select inspection_id as the key property and select Save.

Inspections entity type key

Summary of Entity Types

Entity Type Source Table Key Property
Tickets tickets ticket_id
Inspections inspections inspection_id

All entity types

Step 3: Create Relationship Types

Relationship types represent contextual connections between entity types.

3.1 — Inspections → Tickets (ticket_inspection)

This relationship links an inspection record to its parent trouble ticket.

  1. Select Add relationship from the menu ribbon.

Add relationship

  1. Enter the following relationship details and select Add relationship type:
  2. Relationship type name: ticket_inspection
  3. Source entity type: Inspections
  4. Target entity type: Tickets

Save relationship 3. The Relationship configuration pane opens. Enter the following:

  • Source data: Select your workspace → your Lakehouse → the inspections table. > This table links Inspections and Tickets because it contains the ticket_id column that references the Tickets entity. Relationship Source Data
  • Source entity type > Source column: Select inspection_id > This matches the key property on the Inspections entity.
  • Target entity type > Source column: Select ticket_id > This matches the key property on the Tickets entity (ticket_id in the tickets table). Relationship Source Target Entity Type
  • Select Create. Relationship configuration

Summary of Relationships

Relationship Name Source Entity Target Entity Link Column
ticket_inspection Inspections Tickets ticket_id

Step 4: Validate the Ontology

After creating all entity types and relationships:

  1. Review the configuration canvas — you should see both Tickets and Inspections entity types with a relationship arrow between them.

Final ontology canvas 2. Verify each entity type has:

  • A valid data binding (Bindings tab shows the source table).
  • A key property set.
  • Verify the relationship shows the correct source/target columns.

Step 5: Entity Type Overview

After saving, the ontology takes 15 to 20 minutes to sync and materialize the data. Once ready, you can view the entity type overview for each entity.

  1. Select the Tickets entity type and then click on Entity type overview. Tickets entity type overview
  2. View the dashboard. Tickets entity dashboard
  3. Select the Inspections entity type and then click on Entity type overview. Inspections entity type overview
  4. View the dashboard. Inspections entity dashboard

Note: If the overview still shows "Setting up", wait a 15 to 20 minutes and refresh the page. The ontology needs time to process the data bindings against the Lakehouse tables.

Next Steps

  • Query the ontology using natural language through Fabric IQ.
  • Test the Fabric Data Agent — continue with the next section in the Build solution guide.
  • Enrich the ontology with additional data sources or time-series bindings. See Tutorial Part 2: Enrich the ontology.

Troubleshooting

Issue Resolution
Unable to create ontology item Ensure the Ontology (preview) feature is enabled at the tenant admin level.
Tables not appearing in data source selection Verify that the tables are loaded into the Lakehouse and accessible in OneLake.
Relationship creation fails Confirm both entity types have key properties set before creating the relationship.

← Back to previous page