--- description: >- See your Workspace's semantic layer as a graph in Studio: models and their fields, metrics and glossary terms, and how they connect. --- # Knowledge Graph The Knowledge Graph shows your Workspace's semantic layer as one picture: every model with its dimensions, measures and filters, every metric and every glossary term, and the connections between them. Use it to: - Check what a change could affect before you edit or delete a model. - See which models a metric uses. - Find glossary terms that aren't mapped to anything yet. - Review how glossary terms are organized into a hierarchy. The graph is read-only. You change the semantic layer in the Models, Metrics and Glossary editors. !!! info "Availability" - **Who can do this:** members with the **Edit** capability in the active Workspace. - **Deployment:** Actian SaaS and self-managed deployments, including air-gapped deployments. ## Open the graph Both ways of opening the graph show the same graph. - **Graph tab (main entry point).** On any Studio page, click **Graph** on the right-hand panel rail, next to **Steward**. (**Docs** also appears there, except on air-gapped deployments.) The panel shows a compact version of the graph next to the page you're working on. - **Full page.** Go to `/studio/graph` in your AI Analyst address. The full page adds zoom controls and a minimap. There is no link to it in the Studio navigation. You can also link straight to the Graph tab: add `?panel=graph` to any Studio address. The right-hand panel opens on the **Graph** tab. The right-hand panel remembers whether it's open, which tab is active and its width in your browser between visits. !!! note The right-hand panel and the Studio pages need a window at least 768 pixels wide. On narrower windows, Studio shows a small-screen notice instead. ## What the graph shows The graph covers the Workspace you're in. It includes every model with its dimensions, measures and filters, every metric and every glossary term that hasn't been deleted. ### Node types The legend at the bottom of the graph lists the node types in this order. Each type has its own icon and colour, and the legend colour matches the node. | Legend label | What it is | Colour | | --- | --- | --- | | **Model** | A [model](models/README.md) | Blue | | **Dimension** | A [dimension](models/dimensions.md) on a model | Sky blue | | **Measure** | A [measure](models/measures.md) on a model | Green | | **Filter** | A [pre-defined filter](models/pre-defined-filters.md) on a model | Slate grey | | **Metric** | A [metric](metrics.md) | Violet | | **Term** | A [glossary term](glossary.md) | Amber | Each node shows the entity's name on one line and, if it has one, the first two lines of its description. A glossary term shows its definition as plain text, without formatting. ### Connections Lines between nodes show how entities connect. Lines have no arrows or labels, and the graph has no legend for them, so tell them apart by colour and style: | Connection | From → to | Line | | --- | --- | --- | | Relationship | A model → the **Target model** of each relationship defined on it | Solid blue | | Model field | A model → each of its dimensions, measures and filters | Dashed grey | | Metric to model | A metric → its **Anchor model** and each model it joins | Dashed violet | | Term mapping | A glossary term → each entity it's mapped to on its **Discovery** tab | Solid amber | | Term hierarchy | A glossary term → its **Parent term** | Dotted slate | A metric has one line to each distinct model it uses. Joins are resolved through the relationship they follow. ### Layout The graph is laid out left to right across models, metrics and terms. A model's fields hang in a column under the model: dimensions first, then measures, then filters, each sorted A to Z. ## Work with the graph | To do this | Do this | | --- | --- | | Open an entity | Click its node. Models, metrics and terms open their own page. Dimensions, measures and filters open their page under the parent model. | | Show a model's fields | Fields start collapsed. Click the **+** control under the model, which shows the number of fields (**Expand # model fields**). Click **−** (**Collapse model fields**) to hide them again. Other nodes don't move. | | Hide or show a node type | Click the type in the legend. Its nodes and their connections are hidden, and the type is dimmed in the legend. Click again to show them. All types show by default. | | Re-centre the graph | Click **Center view** (the icon at the right of the legend). It zooms to the highlighted node if there is one, otherwise it fits the whole graph. | | Zoom and move around | Scroll or pinch to zoom between 10% and 150%, and drag the background to move around. The full page also has zoom buttons and a minimap. | You can't drag, connect or edit nodes in the graph. Your type choices and expanded models are not saved. They stay while you move between Studio pages (the panel and the full page each keep their own), and they reset when you reload the browser, leave Studio, switch Workspace or leave the full graph page. ### The Graph tab follows your page In the right-hand panel, the graph follows what you have open: - On a model, metric or glossary term page, that node is outlined and the graph centres on it. - On a model's **Dimensions**, **Measures** or **Filters** tab with a field selected, the field is highlighted and its model's fields expand automatically. - On other Studio pages, the graph highlights the model you're on, or nothing. The full `/studio/graph` page never highlights a node. ## Empty and error states | Message | Meaning | | --- | --- | | "No semantic layer entities yet. Create models, metrics, or glossary terms to populate the graph." | The Workspace has no models, metrics or glossary terms yet. | | "Could not load the knowledge graph." | The graph couldn't be loaded. This also happens if you have Studio access but not the **Edit** capability in the Workspace, for example when you only manage members. | ## What the graph doesn't show - Relationship details such as relationship type, join type or join keys. Open the model's **Relationships** tab to see them. - Deleted entities. Connections to a deleted entity are dropped too. - Which AI Analysts can use which models, metrics or terms. The graph shows the whole semantic layer of the Workspace. - Data sources and AI Analysts. - Links between glossary terms other than parent and child. ## Related - [Searching the semantic layer](search.md) - [Relationships](models/relationships.md) - [Metrics](metrics.md): anchor model and joins - [Glossary](glossary.md): mappings and term hierarchy - [Using Steward](../steward-ai-agent/using-steward.md)