PageRank

PageRank expressed as a population of random surfers on a fixed six-page graph (pages A–F). Each surfer carries a damping taxonomy, and per-page policy tables blend following an outlink with teleporting, weighted by the damping factor. Running three damping values side by side lets the long-run visit frequencies converge toward three different PageRank vectors over the same topology.

Shows a classic iterative graph algorithm emerging from independent agents rather than matrix math. The empirical visit distribution can be checked against the analytic power-iteration solution, making it a clean convergence and correctness benchmark.

# page-rank.osi.yaml — emitted automatically
semantic_model:
  name: "page-rank"
  source: "duckdb://page-rank.db"
  entities:
    - name: surfer
      primary_key: id
  dimensions:
    - name: state
      type: categorical
    - name: t
      type: time
  measures:
    - name: row_count
      agg: count
    - name: active
      agg: sum
      filter: "state = 'ACTIVE'"