Knowledge Graph
Switch to Node view with nothing selected and the main area is the whole graph: your network of concepts.
- A dot = one node
- A line = two nodes appearing in the same note
- Brighter line = they co-occur more often
The graph draws co-occurrence, not link direction
A line does not mean "A links to B"; it means "A and B have been discussed together". So the graph reads as proximity between concepts, not as citation paths between notes.
Two Renderings
Settings → Appearance → Node Graph switches between:
- 2D — a flat layout, best for reading the overall structure and its groupings
- 3D — a rotatable globe, easier to tell layers apart once there are many nodes
Interaction
| Action | macOS | iOS |
|---|---|---|
| See a node's details | Hover and the panel appears | Tap; it rises from the bottom |
| Keep the panel | Click the node to pin it | The panel is a sheet already |
| Rearrange the layout | Drag nodes | Drag nodes |
| Zoom / pan | Trackpad gestures | Pinch / drag |
While you dwell on or pin a node it becomes a slowly rotating sphere, with bands of light travelling out from it along each edge to its neighbours — marking which node is currently the centre in a dense network. The direction is always outward.
What the Info Panel Holds
| Section | Content |
|---|---|
| Overview | Reference count, neighbour count, type badge |
| Co-occurring nodes | Nodes it often appears with, each with a connection strength |
| Similar nodes | Nodes close in meaning that have never co-occurred — loads once the panel is pinned |
| Links | Notes citing it, clickable |
The similar-nodes section deliberately shows no number: the strengths above are facts about links you made, while closeness in meaning is only the model's judgement, and the two should not look alike. It needs semantic relations to be available.
Network Insight
The overview panel summarises the whole network — how many nodes and links it holds, how densely connected and how tightly grouped it is, and how the badges are distributed — and adds one line of insight beneath. Only the single most pertinent observation is shown:
| When | What it says |
|---|---|
| Tightly grouped, with no bridge nodes at all | Echo-chamber risk — clustered and sealed off |
| Several bridge nodes | Key entry points for expanding your boundaries |
| Several nodes newly active | Your focus may be shifting |
| Some nodes connected to nothing | Suggests connecting one to existing concepts |
| Plenty of nodes but few links between them | Associations still undiscovered |
If none apply, nothing is shown — a healthy graph does not need commentary.
This line appears on the overview panel only, never on a single node's panel: every other panel answers "how is this node doing", and only the overview speaks about the network as a whole.
The Graph Sees the Current Project
The graph, the co-occurrence matrix and the network summary all cover the currently selected project. This is why archived notes should not be moved into an "archive project" — it would take them out of your graph. See PARA.
Next Steps
- Co-occurrence Matrix — The same relationships as a grid
- Link Predictions — The lines still missing
- Node Types — What the badges mean