Follow the Money

This repository contains a pragmatic data model for the entities most commonly used in investigative reporting: people, companies, assets, payments, court cases, etc.

The purpose of this is not to model reality in an ideal data model, but rather to have a working data structure for researchers.

followthemoney also contains code used to validate and normalize many of the elements of data, and to map tabular data into the model.

Documentation

For a general introduction to followthemoney, check the main documentation:

• https://docs.alephdata.org/developers/followthemoney

Part of this package is a command-line tool that can be used to process and transform data in various ways. This is documented here:

• https://docs.alephdata.org/developers/ftm

Releasing

We release a lot of version of followthemoney because even small changes to the code base require a pypi release to begin being used in aleph. To this end, here's the steps for making a release:

git pull --rebase
make test
bumpversion patch
git push --tags

This will create a new patch release and upload a distribution of it. If the changes are more significant, you can run bumpversion with the minor or major arguments.

Docs

When the schema is updated, please update the docs, ideally including the diagrams.

For the RDF namespace run make namespace.

Namespace URI is https://w3id.org/ftm/. Docs are auto-hosted on github pages, and redirects are managed at https://github.com/perma-id/w3id.org/

Command-line utility

This package installs ftm, a command line utility for working with followthemoney-formatted entity graph data.

Input and output stream types

ftm is based on the idea of streamed graph data that is piped between small sub-commands. Unfortunately, there is not one single data format that is appropriate for all stages of the utility. Instead, the following formats are used:

• Entity streams. An iterator of line-based JSON representations of followthemoney entities.
• Result streams. A set of results from an enrichment process. Each JSON-formatted line may include multiple entities, one of which is the entity being enriched (subject), the main match entity (candidate) and other entities that the remote API may have returned.
• Recon files. These files define a mapping from a source entity ID to a canonical entity ID, e.g. after a de-duplication process has been applied.

Other commands

• export-cypher - Dump a Neo4j OpenCypher script.
• export-excel - Dump an Excel file, one sheet per schema.
• export-gexf - Dump a GEXF graph file.
• apply-recon - Read a recon file and re-write entity stream to their canonical ID. Should be followed by aggregate.
• filter-results - Read a recon file and emit only results where candidate and subject are confirmed matches.