The knowledge graph

The knowledge graph at /knowledge is an interactive canvas that draws every document and extracted entity you have filed, with edges showing the relationships the pipeline found between them. It is the main way to explore your library spatially.

Clusters

Documents are automatically grouped into clusters by type — for example textbook, web, lab, project, docs, and other. Each cluster appears as a single labelled circle on the canvas. The number inside the circle tells you how many items are inside.

You do not create or name clusters manually; the pipeline assigns categories during ingest.

Opening and folding territories

Click a cluster circle to expand it into a territory — a soft dashed ellipse that reveals the individual document nodes inside. The simulation reheats and resettles the layout each time you expand or fold.

To fold a territory back, click the cluster circle again, or click anywhere on the blank canvas outside all open territories. Clicking outside collapses all open territories at once and re-frames the view.

You can pan by dragging, zoom with the scroll wheel or trackpad, and use the +, , and buttons in the bottom-right corner. The button fits the whole graph into the viewport.

Individual document nodes can be dragged to rearrange them within an open territory.

Hover to light a neighborhood

Hovering over any node dims everything except that node and its direct connections. A peek card appears in the bottom-left showing the node's title, topics, and connection count.

Hovering over a cluster node instead lights the cluster and all the document nodes currently visible inside it.

Node size and what it means

A document node's size is proportional to how many other nodes it connects to — more connections, bigger circle. Highly connected documents are the conceptual hubs of your library. Entity nodes (concepts, people, places) follow the same rule.

Reading a document inline

Click a document node (when the cluster is expanded) to select it. A detail panel slides in showing the document's title, topics, and linked nodes. Click Read here to open the inline reader — the graph shrinks to the left half of the screen while the document content appears on the right. Close the reader to return to the full canvas.

Tip

You can also navigate to a specific document node via a deep link: /knowledge?node=<id>. The graph will expand the right cluster and open the inline reader automatically. The processing card on the Upload page links here once your document is ready.

Search →