Graph Structural Gap Detectionbacklog_item

visualizationaiwebgraph
1 min read · Edit on Pyrite

Identify clusters of entries that are topically related but not yet linked — "structural gaps" — and surface them as suggested connections or research prompts.

  • After community detection, find pairs of clusters with high intra-cluster density but low inter-cluster connectivity
  • Highlight gap edges (potential missing links) as dashed lines on the graph
  • Show a "Gaps" panel listing cluster pairs with suggested bridge entries
  • Optionally use AI to propose why two clusters might connect and suggest new entries or links
  • Complements the existing "wanted pages" feature by finding missing relationships, not just missing pages
  • Inspired by InfraNodus (Paranyushkin, WWW'19) which detects structural gaps between discourse communities to surface blind spots and prompt new insight.