Kursübersicht
AcademyModul 3 · Entities up close
The entity table: your graph as data
Das lernst du
- You can open the entity table with the titlebar button, ⌘⇧L, or an "Open as table" deep link.
- You can point the table at any set: all entities, one type, "Type unknown", a collection, or a case.
- You can explain what the "Connections" and "Notes" columns each count, and why they differ.
- You can filter and sort the table, then hand the visible rows to the graph with "Show on canvas".
- You can open any entity's 360° page straight from its row.
Tend your knowledge graph for a few months and it turns into a constellation: hundreds of dots and clusters. Great for seeing shape, bad at a plain question like "which people do I write about the most?" You could hover over the dots one by one, or ask for the whole graph as a spreadsheet.
That's the entity table: a spreadsheet-style view of one set of entities from your graph, sortable and filterable however you like. Same data as the canvas, browsed as rows and columns instead of dots and lines. Each row is one entity; click it to land on that entity's full 360° page.
Opening the table
The table lives in a tab titled "Entities", and there are several ways in:
- The titlebar has a table-icon button labeled "Entity table", just left of the graph button. It stays highlighted while the table is your active view.
- The keyboard shortcut is ⌘⇧L (Ctrl+Shift+L on Windows/Linux). You can rebind it in Settings ▸ Shortcuts.
- In global search (⌘⇧F), run the navigation action "Entity table".
- "Open as table" deep links turn up in a few spots: every type card in Knowledge graph tuning ▸ "Entity types", each saved collection in the graph canvas's Collections FAB, and a case file's detail view. Each one opens the table already pointed at that set.
- The workspace empty-state launcher also has an "Entity table" card with its shortcut.
Choosing what to look at
The source switcher in the top-left names the set you're looking at. Click it to open a menu and retarget the table. Every entry carries a live count, so the menu doubles as a quick census of your graph.
| Source | What the table shows |
|---|---|
| "All entities" | Everything in the graph, with a running total. |
| A type (under "Types") | Every entity of one kind: Person, Organization, Place, and so on. Only types that actually have members show up, each with its colored dot and count. |
| "Type unknown" | Entities the AI hasn't typed yet. It shows up whenever some entities are still untyped, ready for you to tidy up. |
| A collection (under "Collections") | A saved living set. Its recipe re-runs against the current graph, so the member count is always current. The group hides itself when you have no collections saved. |
| A case | The entities pinned to a case file, opened from the case's "Open as table" button. The table treats them like a hand-picked collection. |
- Entity
- A thing the AI extractor spotted in your notes, like a person, a project, or an idea. It's stored as a node in the knowledge graph. One note can mention many entities, and one entity can turn up in many notes.
- Entity set
- Whatever the table is showing right now: all entities, one type, "Type unknown", a collection, or a case's pinned entities.
- Connections
- How many distinct entities this one shares a drawn graph edge with. These are exactly the lines you'd see on the canvas.
- Notes
- How many notes mention the entity. It's the backlink count, a rough measure of how prominent the entity is.
- Collection
- A living set saved as a recipe of steps. It re-evaluates against the current graph every time it renders.
Four columns, one sorting rule
| Column | What it shows | First click sorts |
|---|---|---|
| "Name" | The entity's display label, in bold. Long names get truncated, with the full name on hover. | A → Z |
| "Type" | A kind chip with the type's fixed color dot. Untyped entities show a dash. | A → Z |
| "Connections" | Distinct entities this one shares a drawn graph edge with. | Highest first |
| "Notes" | How many notes mention this entity. | Highest first |
The default sort is Notes, descending, so your most-mentioned entities sit at the top the moment the tab opens. Click any header to sort by it, click again to flip direction. Ties break alphabetically by name, so the order never jitters.
It stays fast at any size: the first 500 rows draw right away, and a "Show all (+N)" bar loads the rest when you want them.
Kurzer Check
An entity shows Notes 42 but Connections 3. What does that tell you?
Filter, then hand off to the canvas
The "Filter by name…" box does a case-insensitive substring match on the Name column, and the "{n} entities" caption next to the source switcher ticks down live as the set shrinks. Read that label literally: it matches *names only*, not aliases, types, or note titles.
Once you've narrowed the table down to the rows you care about, "Show on canvas" (top right) projects exactly those visible rows onto the graph canvas as the focused set, then flips you over to the graph tab. That's the table-to-picture hand-off: filter by the data, then go study the shape. The button goes disabled when no rows match.
Kurzer Check
You filtered "All entities" from 300 rows down to 12, then clicked "Show on canvas". What does the graph show?
From a row to a 360° page
Every row is clickable. The click opens that entity's 360° page as a regular tab: its ranked connections, aliases, and every note that mentions it. You can keep several entity pages open side by side, but reopening the same entity by any of its aliases just focuses the tab already open.
That's the table, the census view of your graph. Next up is where every row leads: the entity 360° page, with the source sentence behind every connection.