Select data for best performance and risk typology coverage

The primary functionality of AML AI is to support risk typologies related to transaction monitoring. These typologies include the following:

  • Funneling (placement)
  • Pass-through funds
  • Structuring
  • High-risk jurisdictions
  • Cross-border activities
  • Shell companies
  • Professional enablers
  • Money mules

To optimize coverage, you should not only provide MANDATORY fields but also RECOMMENDED fields because some of them enable additional features that are critical risk indicators for the less common typologies.

Data fields categorized as RECOMMENDED can improve risk typology coverage in two ways:

  • By supporting less common typologies that don't have any supporting features calculated from the MANDATORY data fields (for example, money laundering through high-risk jurisdictions)
  • By strengthening the coverage for an already supported typology with new features that yield additional results (for example, money muling)

The following table summarizes the purpose of all RECOMMENDED fields in the AML AI schema.

FieldTablesPerformance impact?Typology coverage impact?Other uses
nationalitiesPartyYes Yes - money muling
  • Money laundering through shell companies and professional enablers
  • Money laundering through high-risk jurisdictions and cross-border activities
birth_datePartyNoNo Field can be used for your own fairness analysis. Field can be used in feature generation, depending on engine version.
establishment_datePartyNoNo Field can be used in feature generation, depending on engine version.
genderPartyNoNoField can be used for your own fairness analysis.
is_entity_deletedNoNoField can be necessary to correctly model how entities change over time, depending on how you manage your data internally (see Understanding how data changes over time).
source_systemNoNoField helps you to manage dataset quality.