Visualisation & analytics

Visualisation mode renders your map analytically. Switching to it (and between lenses) never changes your data — only how it is drawn. Switch modes with the Edit/Visualise/Table toggle next to the map name.

Unlike Edit mode, visualisation mode can show every node in the map, whether or not it’s been placed on the Edit canvas yet — the force layout positions nodes itself, so there’s nothing to place first. This is controlled by the Visualise unplaced nodes toggle in the sidebar’s Settings panel, and is on by default.

Lenses

• Landscape (force-directed topology) — the overall shape of the network, laid out by connection strength.
• Bridges (betweenness centrality) — highlights the nodes that act as bridges between parts of the network. For large maps this is computed in the background; a “Computing centrality…” indicator shows while it runs.
• Capabilities (field clustering) — nodes coloured, sized, and clustered by a chosen field, e.g. a capability.

Filtering

The Filter panel is a top-level sidebar tab (next to Visualisation), separate from Weighting — pick which connection type(s) stay in view. Filtering is orthogonal to weighting: applying a filter never changes or discards your Weighting panel setup (Presence/Count/Score), it just narrows which connections that setup applies to. Filtering to no types (the default) shows every connection.

Filtering only ever changes what’s drawn — connections outside the filter fade out rather than disappear, and any node left with no visible connections dims in place rather than being removed. The force layout itself never reacts to a filter: node positions are computed from the whole (unfiltered) network and stay fixed as you filter, so nothing jumps or re-settles.

Connection weighting

The Weighting panel controls how connection weight is computed — it’s shared across lenses, not tucked inside Landscape’s or Bridges’ own controls, because it governs both: Landscape’s layout (a heavier pair sits closer together) and Bridges’ path costs alike.

• Presence (default) — any connection between a pair counts, zero configuration required.
• Count — a weighted count of connections between a pair, scaled across the whole graph.
• Score — assign each connection type a score from 0 to 1; a pair’s weight sums the scores of the types connecting it.

A connection with no assigned field (just a Reason) has no “type” for Score to key off, so it’s treated per mode: it counts normally under Presence and Count, and contributes nothing under Score.

Directional highlighting

When the Filter panel has exactly one directed connection type selected and you select a node, the canvas colour-codes the selected node’s neighbourhood instead of the usual dim-the-rest highlight: the selected node itself, nodes that point at it, nodes it points at, and nodes with a link in both directions each get a distinct colour. A legend in the bottom-left corner of the canvas names the field and its reverse reading (e.g. “funds” / “funded by”) so you can tell which colour means which direction.

Encodings

You can size and colour nodes by a node field — the Appearance panel handles this independently of which lens is active (betweenness centrality always drives size in its own lens; field clustering always drives colour in its own lens). Whenever a colour encoding is active, a legend in the bottom-right corner of the canvas explains it — swatches and values for a categorical field or clustering, a gradient with its low/high values for a numeric field. Connections have no colour of their own — their width encodes weight instead, noted at the bottom of the same legend. Click a node or connection to see a read-only detail panel on the right, which highlights the node’s immediate neighbours, dimming the rest. Hovering a node shows a pointer cursor and a highlight ring.

The panel opens the moment you click anything, showing a prompt to select a node or connection before that. Close it with its ×, the Escape key, or by clicking empty canvas — it then collapses to a slim labelled edge on the right; clicking that edge (or selecting something new) reopens it.

A selected node’s detail panel lists its own connections as clickable rows — each names the connection type and the other node, and clicking either jumps the selection straight there, so you can navigate the network from panel to panel without hunting for the next node on the canvas.

Settings

The Settings panel holds display options that aren’t specific to a lens or appearance overlay — currently just Visualise unplaced nodes. It stays interactive even when the current lens has nothing to draw, so you can always switch it back on to see why.

Saving views

Save a view to capture the current lens, weighting mode, encodings, and viewport (zoom/pan) so you can return to it later from the toolbar. A saved view is metadata only — it never changes or duplicates the underlying map data.

Before any view exists, there’s just one button: Save view. Once you’ve saved one, two save actions sit side by side, and neither is the “advanced” one:

• Save overwrites whichever view is currently loaded. A map you only ever keep one view of never needs anything else.
• New view always creates an additional view, leaving whatever’s currently loaded untouched. This is how you build up a set of views on one map.

Neither prompts for a name — the first view on a map is called “default”, and later ones “view 2”, “view 3”, and so on, and a newly-created view becomes the active (loaded) one straight away.

With only one saved view on a map, the toolbar shows nothing else for it at all — no name, no picker, no rename, no delete. It’s always kept loaded, so there’s nothing to pick between and no separate step to load it, even right after deleting down to it from a second view. Once a second view exists, a dropdown appears to choose between them, with a pencil/trash pair next to it: the pencil turns the dropdown into a text box in place, ready to rename; the trash deletes the loaded view (a confirmation dialog checks first — this can’t be undone).

Whichever view was last loaded is remembered, per map and per browser, so reopening a map picks up where you left off rather than starting from “no view loaded” every time.