What is the Knowledge Graph?

This page visualises the connections between three folklore classification systems as an interactive force-directed network. Each dot is a node and each line is a relationship discovered in the RunoVerse corpus data. The graph reveals how international tale types, narrative motifs, and Estonian treasure legends interrelate.

Node Types

• Blue nodes — ATU Tale Types from the Aarne-Thompson-Uther international folktale catalogue (e.g. ATU 301 "The Three Stolen Princesses"). Larger circles indicate more connections.
• Green nodes — TMI Motifs from the Thompson Motif Index, representing recurring narrative elements (e.g. D1810 "Magic knowledge"). These are smaller dots since there are many of them.
• Orange nodes — Treasure Legends from the Estonian varandusemuistendid corpus (6,196 legends across 93 motif types). These are the largest dots.

Edge Types (Lines)

• has_motif — an ATU tale type contains a particular TMI motif
• combines_with — two legend motif types frequently co-occur within the same narrative
• classified_as — a treasure legend maps to an ATU tale type

Interaction

• Click a node to select it. The detail panel at the bottom shows its name, type, and all connections. Connected nodes are highlighted while the rest dim.
• Drag a node to reposition it within the layout.
• Scroll wheel to zoom in/out (0.2x to 5x range).
• Double-click the canvas background to deselect and restore the full view.
• In the detail panel, click any connection tag to jump to that node. Use the "Open in explorer" link to view the full entry on the dedicated ATU, TMI, or Legends page.

Filter Buttons (Left Sidebar)

• All — show every node and edge
• ATU Types — show only blue ATU nodes and their edges
• TMI Motifs — show only green TMI nodes and their edges
• Legends — show only orange legend nodes and their edges

Filtering is useful for dense graphs. When a node is selected and a filter is active, cross-type connections of the selected node remain visible even if those neighbour nodes do not match the filter.

Tips

• The initial layout is computed with 200 iterations of force simulation. Clusters of tightly connected nodes appear close together.
• If the graph feels crowded, zoom in to a region of interest or use the type filters to reduce clutter.
• Node size reflects type: legends are largest (8 px), ATU types medium (6 px), and TMI motifs smallest (4 px). Selected nodes double in size; hovered nodes grow by 1.5x.