Methodology — Network Mapping
Last updated: October 21, 2026
A network map shows people, companies and institutions as nodes and the documented relationships between them as edges: a board seat, a donation, a contract, a shared address. A good map makes a pattern visible that a list of facts hides. A bad one makes coincidence look like conspiracy. Every rule below exists to keep a map on the right side of that line.
1. Schema before volume
Define node and edge types once; avoid “everything connects to everything” sprawl.
You do not need to invent a schema. FollowTheMoney (FtM), the data model used by the OpenAleph document platform and by yente, a sanctions-screening engine, defines entity types for financial-crime investigations, such as people and companies, and lets entities reference each other to form a graph. It also models the documents used as evidence. Using its types makes your data exchangeable with those tools.
2. Edge evidence rule
Every edge has a citation column before visualization.
An edge also needs a date range. “Served on the board” is a different claim in 2015 than in 2025.
3. Iterative expansion
Start from seed set (known persons/orgs). Expand one hop; review; repeat. Stop when marginal edges are noise.
Before each expansion, merge duplicates. The same company spelled three ways becomes three nodes and hides real links. OpenRefine’s clustering feature finds and merges near-identical values in a spreadsheet before you import it.
4. Layout honesty
Force-directed layouts are not proof of clusters being “teams.” Label clusters descriptively.
Analysis software can compute measures such as community detection and centrality. Gephi, a free open-source network tool, includes built-in algorithms for both and says it handles networks from ten to ten million nodes. These measures describe the shape of your data. They inherit every gap in your sourcing, so a “central” node may only be the best-documented one.
5. Export for audit
CSV/JSON of nodes and edges for peer review prior to publication.
Publish or retain the edge list with its citations, so a reader or editor can check any line on the map against its source.
Related skills
network-analysis-specialist, corporate-intelligence-investigator
See also Methodology — Corporate and Financial Investigation.
Sources
- FollowTheMoney data model (read 2026-10-21)
- Gephi (read 2026-10-21)
- OpenRefine (read 2026-10-21)
Expanded with AI research and drafting assistance from the primary sources cited; editorial review pending.
