The Knowledge Graph Explorer
Every table in OpenMetadata carries a Knowledge Graph tab that renders its neighborhood as an interactive graph. It is the fastest way to see, rather than query, how one asset sits in the wider catalog.
The tab is hidden unless RDF is enabled, and is marked Beta. It reads from
GET /api/v1/rdf/graph/explore, which is admin-only.The Toolbar
The zoom controls on the canvas zoom in and out, fit the graph to the screen, and refresh it.
Modes
The mode switch changes what the graph is about.Ontology Mode
Ontology mode hides the technical plumbing and draws the business model as applied to real data. Concepts are drawn as nodes, connected by typed relations such as Broader, Narrower, and Related to. Assets, columns, and domains that are mapped to a concept connect to it through Has glossary term. A concept that carries a mapping is marked Mapped to business ontology.
- A Business concept selector appears next to Level to choose which concept the graph is centered on.
- The footer drawer shows Properties, Axioms, and Coverage in place of Columns, Relationships, and Gaps.
Properties
Lists the declared and inherited properties of the concepts in view, with each property’s Range, Domain concept, and Cardinality. Cardinality is shown only when the ontology declares it.
Axioms
Lists every relationship in view, ungrouped, with its Subject, Predicate, Object, and Family. Filter the list by family, for example Ontology or Governance.
Coverage
Lists assets missing a glossary term in the current scope, with where each asset lives, what is missing, and a suggested fix. The number next to Coverage in the footer is the count of unmapped assets.
Levels
The Level control sets how far the graph reaches from the asset you opened it from. Each option shows its number and name in the control, for example 2 · Direct.
The graph is laid out so that each level sits in its own band, and the caption in the footer reports the current level, for example “Level 2 of 3”. Raising the level adds an outer ring without moving the entities already on screen.
Filtering
Entity Type and Relationship Type are multi-select filters. Each opens a searchable list with a count next to every option, a Select all checkbox, and Clear all. The button shows how many options are selected.
- Entity Type groups entity types into categories such as Governance, Owners, and Other. Expand a category to pick individual types such as Column or Domain.
- Relationship Type groups relationships into families such as Ontology, Governance, Ownership, and Other. Expand a family to pick individual relationships such as Contains.

Reading the Graph
These are the visual cues that show what each node and edge means.Relationships
Every relationship keeps its own predicate and direction, drawn as a labeled edge with an arrow. When the same predicate repeats, for example a table with seven columns, the edges are bundled into one edge with a count, such as Has column ×7.Relationship Families
The color and dash pattern of an edge show its family. The family button (for example 5 families) opens a legend with the families present in the current view and how many relationships each holds.
Select a family in the legend to highlight it and dim the rest of the graph.
Grouped Nodes
When many entities of one type connect to the same asset, the graph groups them into a single node with a count badge, such as Data Assets (3) or Columns (7). A group lists the first few members and a +N more label.- Select Expand group to draw every member as its own node.
- Select Collapse groups in the footer to bundle them again.
Selecting Nodes and Edges
Selecting a node or an edge opens an inspector with its details, including its type and its relationships. For an asset, the inspector also has actions to preview the asset or open its page.View Settings
The gear icon opens the View menu.
The footer repeats the current settings, for example “Balanced · repeats bundled”, how many relationships are individual and how many are bundled, and the label mode.
The Details Drawer
The footer of the graph has a row of tabs. Select one to open a table beneath the graph, and select the close button to dismiss it. Each table has a search box.Columns
Lists the columns of the selected table with their type, glossary terms, and classification tags. Filter the list by All or Unknown, where Unknown are columns with no glossary term.
Relationships
Lists every relationship in view, ungrouped, with its Subject, Predicate, Object, and Family. Filter the list by family.
Gaps
Lists assets in the current scope that have no glossary term, with where each asset lives and a Map to a glossary term action. The number next to Gaps in the footer is the count of unmapped assets.
Status Messages
A truncated graph is labeled rather than silently misleading.
Troubleshooting
The tab is not there
The tab is not there
RDF is disabled on this server, or you are not looking at a table. The tab is registered for tables and hidden when RDF is not enabled.
No knowledge graph data available
No knowledge graph data available
RDF is enabled but the graph has not been populated. Run the RDF Knowledge Graph Indexing application from Settings → Applications.
The graph is dense and hard to read
The graph is dense and hard to read
Drop Level to 1 · Entity or 2 · Direct, keep Presentation on Balanced, filter by Entity Type or Relationship Type, or set Relationship labels to No labels.
Exporting the Graph
Export Graph in the View menu downloads the current view as PNG, JSON-LD, Turtle, or CSV. The export follows what is currently displayed. In Ontology mode, the PNG exports the view, while the other formats export the underlying entities and relationships. To take the data with you from a script, use the export endpoint the tab reads from:format accepts turtle or jsonld. entityTypes and relationshipTypes accept comma-separated filters.
Next
Querying with SPARQL
When clicking is not enough.
Graph Insights
Importance, communities, and paths computed over the whole graph.