Last updated:
Entity Graph
Paste a list of relationships — who contacts whom, which account links to which — and see the network. The lab uses real networkx to compute degree, betweenness, and eigenvector centrality, find connected components, and detect communities, then draws an interactive force-directed graph. Betweenness highlights the brokers that bridge clusters. The engine is Python (networkx) compiled to WebAssembly and runs entirely in your browser — your data is never uploaded.
② Edge list
Reading a link-analysis graph
In an entity graph, nodes are people, accounts, or assets and edges are the relationships between them. Degree counts direct connections; betweenness centrality measures how often a node sits on the shortest path between others, so high-betweenness nodes are the brokers whose removal would fragment the network. Community detection groups densely-connected nodes, surfacing clusters that may correspond to teams, cells, or shared infrastructure. Building this from an edge list normally means a Python notebook or paid link-analysis software; here it runs on networkx in your browser, so the relationships you are mapping never leave your device.
Frequently asked questions
What input format does it take?
One relationship per line: “A -> B”, “A,B”, a tab, or two-or-more spaces between names. An optional third field is treated as edge weight. Lines starting with # are ignored. You can also load a CSV edge list.
What does betweenness centrality tell me?
It identifies brokers — nodes that lie on many shortest paths between others. In an investigation these are often the key connectors linking otherwise separate groups, and the graph sizes nodes by this score.
Is my data uploaded?
No. The graph is built with networkx compiled to WebAssembly and runs in your browser. After the one-time engine download it works offline and your edge list never leaves your device.
How large a graph works well?
The layout runs in your browser, so up to a few hundred nodes is smooth. Larger graphs still compute their metrics but the drawing becomes dense; for very large networks a desktop tool is better.