What is this?

The Song Type Network is a force-directed graph that visualizes how Finnish runosong types relate to one another through co-occurrence. Two song types are connected when the same poem has been classified under both types — for example, a poem tagged as both “Lemminkäisen surma” (Death of Lemminkäinen) and “Tuonelassa käynti” (Visit to the Underworld). The more poems share both types, the stronger the connection.

Nodes and their size

Each circle (node) represents a single song type. The size of the node reflects the total number of poems classified under that type — larger nodes are more common types in the corpus. The color indicates the thematic category (incantations, epic, lullabies, wedding songs, etc.), as shown in the legend.

Edges and their thickness

Each line (edge) connecting two nodes indicates that at least some poems carry both type labels. The thickness of the line is proportional to the number of co-occurring poems: thicker edges mean stronger associations. Nodes with stronger connections are also drawn closer together by the force simulation, so tightly-clustered groups indicate song types that frequently co-occur.

Filters

Min poems controls which song types appear — setting it higher removes rare types and simplifies the graph. Min co-occurrences controls which connections are shown — higher values keep only the strongest links. Only types that have at least one qualifying connection are displayed (isolated nodes are hidden).

Interacting with the graph

Hover over a node to see its name, poem count, and thematic category. Click a node to enter ego network mode, which shows only that type and its direct neighbors (plus any connections among those neighbors). Drag any node to rearrange the layout manually. Scroll to zoom in or out, and pan by clicking and dragging the background. Use the search box to find and focus on a specific type by name.

What patterns to look for

Dense clusters reveal thematically linked song families — for example, incantation types (purple) often form a tight group because the same poem may contain multiple healing charms. Bridge nodes sitting between clusters are types that connect otherwise separate traditions. Isolated pairs at the graph periphery represent niche co-occurrences. Toggle Labels on/off to reduce visual clutter when exploring dense regions.