Ontology (preview) tutorial part 3: Explore the ontology

In this tutorial step, explore your ontology in different ways. View the ontology in multiple canvas views, explore its entity instances, and explore the ontology's integrated graph.

Explore canvas views

The home configuration canvas of ontology contains three views: full ontology, lineage, and relationship.

Full ontology

The Full ontology view of the canvas shows all entity types and relationships present in the ontology. Select an entity type from the Explorer to center it in the canvas window.

Lineage

The Lineage view of the canvas shows the inheritance structure of entity items. Select an entity type from the Explorer to show it in the canvas along with any inheritance relationships.

Relationship

The Relationship view of the canvas shows the relationships of entity items. Select an entity type from the Explorer to show it in the canvas along with any modeled relationships.

Explore entity instances

When you bound data to your entity types in part 1 of the tutorial, ontology automatically created instances of those entity types that map to the source data rows. View the entity instances with these steps:

1. Start in the Home configuration canvas of ontology. Select an entity type, and View Entity Type details from the top ribbon.

   

2. Open the Instances tab. You see a list of entity instances with data populated from the original lakehouse table data source. For example, the Inventory entity type instances have columns like ProductId, OnHandQuantity, and ShelfAvailableUnits.

   

   Tip

   If data bindings don't load, confirm that the source data tables exist with matching column names, and that your Fabric identity has data access.

Explore ontology graph

Ontology comes with an optional built-in instance of graph in Microsoft Fabric. This feature isn't required for your ontology to work, but you can set it up to add another option for viewing and exploring your ontology details.

This section shows how to create an optional materialized graph from your existing ontology by selecting eligible entities and relationships and materializing the projection. Then, utilize the materialized graph to explore and discover connected data insights.

Prepare the graph

1. Prepare data for graph visualization by defining an entity type key for each entity type. An entity type key shows which field uniquely identifies a row of sample data.

   Note

   Entity types must have entity type keys defined to be eligible for graph projection.

   1. Select an entity type in the canvas and select View Entity Type details.

   2. Select Define entity type key.

      

   3. Select the right key from the following table. Select Save.

      Entity type	Entity type key
      Location	LocationId
      Store	LocationId
      Distribution Center	LocationId
      Product	ProductId
      Frozen Product	ProductId
      Perishable Product	ProductId
      Sale	skip
      Inventory	StoreId, ProductId (multi-select)
      Refrigeration Unit	UnitId
      Shipment	ShipmentId
      Supplier	SupplierId
      Refrigeration Telemetry	skip

   4. Repeat for each entity type until all entity types have keys.

2. Set up the graph instance. From the home canvas ribbon, select Manage graph.

   

3. The Configure Graph page loads with the default selection of Use the entire Ontology. You can see all entity types in the Entities section, and the Preview section shows what the graph looks like. Each entity type is projected as a graph node, and each relationship type is projected as a graph edge.

   

4. Notice that Sale and Refrigeration_Telemetry entity types can't be added to the graph. The graph in ontology doesn't currently support semantic model and eventhouse data sources.

   Verify that all entity types are selected except Sale and Refrigeration_Telemetry. Select Continue.

   Note

   Currently, only delta tables from lakehouses or mirrored databases are supported data sources for the graph view.

5. On the Projection Summary page, you see 10 entities. Select Materialize. Creating the graph model might take several minutes.

   

6. When the graph finishes provisioning, return to the home canvas view.

Explore the graph

1. From the top ribbon, select Explore graph. This button is visible now that the graph is materialized.

   

2. The graph queryset view opens. On the right side of the queryset, select the puzzle piece icon to expand the Components pane. The Components pane shows the list of nodes and edges that are available in your graph.

   

3. Under Nodes, select Store, Product, and Shipment. Under Edges, select Shipment_Store and Shipment_Product. The nodes and relationships are added to the canvas.

   

4. Select Path query from the top ribbon and confirm the Switch when prompted. In this view, you can create complex path-finding queries.

5. Enter the following query details to search for nodes within two hops of the Seattle store's relationship with the supplier Cascade Fresh Products.

   • Start node: Store
   • End node: Supplier
   • Filter start node: City = Seattle
   • Filter end node: SupplierName = Cascade Fresh Products
   • Max hops: 2

   

6. Select Run to run the query and render the result in both the query result canvas and the path results pane.

   

If you want, try more queries and graph features to explore the ontology further before continuing with the tutorial.

Next steps

In this step, you explored the ontology, including the canvas, entity type instances, and graph view. Next, use the ontology agent to explore the data further with natural language queries.

Continue to Consume ontology from agents.