Explore how entities relate, reveal communities, and measure centrality—without uploading private data.
Explore an induced neighborhood inside the current filters. Direction is ignored for neighborhood discovery.
Community labels are deterministic estimates. Filtering changes the analytics shown on the right.
Add a node or reset your filters.
Interpretation. PageRank is calculated with a weighted random-walk model. Detected communities use weighted label propagation; they are not guaranteed to be optimal or stable across edited graphs.
PageRank and degrees are computed on the filtered graph. Node size represents visible degree.
Exact unweighted betweenness, normalized harmonic closeness, and weak bridges/cut vertices. This can take longer on dense graphs.
Import entities and connections, then inspect connectivity, strong links, communities, local triangles and centrality. Useful for organizational diagrams, dependency mapping, research networks and graph-learning workflows.
Your graph and imports are processed in the browser. Browser storage is best-effort; export the editable JSON file for reliable backups. This is not a production graph database and does not use a server.
Create or import a relationship graph, explore node neighborhoods and layout options, and review baseline or worker-backed graph metrics.
Create three nodes labelled A, B, and C, then add directed edges A to B and B to C. Inspect the neighborhood around B and compare a shortest-hop route with the strength-weighted route before exporting a JSON project.
Metrics describe the graph data you supplied; they do not establish identity, causation, or fraud. The graph is limited to 400 nodes and 1,600 edges. Bridge analysis uses an undirected projection. Inverse-strength routes minimize the sum of 1/strength, not physical distance or maximum flow. Advanced analysis uses a Worker and browser storage; real-origin Worker and storage behavior are unverified. Keep a downloaded JSON backup for recovery.
No. Shortest-hop routes count graph edges. The strength-weighted route uses the documented inverse-strength cost, which is not geographic distance or maximum flow.
No. They summarize the structure and values in the supplied graph and should not be treated as evidence of identity, intent, or causation.
