Create theme co-occurrence network visualization — create_theme

Builds a network graph where nodes are themes and edges represent co-occurrence strength (entries assigned to both themes). Requires multi-label assignment columns (theme_membership_*).

Usage

create_theme_network(
  data,
  theme_set,
  output_path = "theme_network.png",
  min_cooccurrence = 3,
  methodology_mode = NULL,
  run_id = NULL,
  max_inline_themes = 30L
)

Arguments

data: Tibble with theme_membership_* columns
theme_set: ThemeSet object
output_path: File path for PNG output
min_cooccurrence: Minimum co-occurrence count to draw an edge (default 3)
methodology_mode: Optional character. When supplied, adds a footer caption identifying the mode + run.
run_id: Optional character: run identifier.
max_inline_themes: Integer; when the graph has more nodes than this after isolated-vertex removal, the network is filtered to the top-N by weighted degree (sum of edge weights). Default 30L.

Value

Invisible adjacency matrix, or NULL if igraph unavailable

Details

at scale the unfiltered network was an unreadable hairball (a large run plotted 417 themes at once with no legend). The max_inline_themes parameter caps the visible network at the top-N most-connected themes (ranked by weighted degree) and adds an inline legend explaining node size + edge width encoding.