ESProfiler Handbook
Models

Relationship Definition Model (RDM)

The Relationship Definition Model (RDM) defines a standardized structure for expressing relationships between entities and framework nodes. It creates a consistent vocabulary for mapping capabilities, controls, or intelligence to taxonomic frameworks.

🎯 Purpose & Core Concepts

An RDM defines a class of relationships that can exist within a given framework. It specifies:

  • The mapping type (e.g., PRODUCT, FEATURE, CONTROL, INTEL)
  • The possible relationship traits that can be assigned
  • Optional modifiers that allow for more granular variations of a relationship

πŸ”‘ Each RDM is identified by its rdm value, which must be unique across all RDMs, not just within its framework. An RDM belongs to one framework (via root) and is created against one mapping type.


🧩 RDM Structure

ComponentDescriptionConstraints
modelDeclares the model definitionRDM: followed by a semantic version, e.g. RDM:1.0.0
rootRoot of the framework the RDM applies toMust match ^[A-Z0-9-]{2,30}$
createdOnDate the RDM was createdISO 8601 timestamp
modifiedOnDate the RDM was last modifiedISO 8601 timestamp
rdmUnique identifier for the RDMMust match ^[A-Z0-9-]{2,30}$
typeMapping type this model applies toOne of: VENDOR, PRODUCT, FEATURE, CONTROL, INTEL, GENERIC
nameHuman-readable name3-100 characters
descriptionDetailed explanation of the RDMRequired, up to 10,000 characters
relationsSet of relationship traits1-20 traits
modifiersOptional variations of relationships0-5 modifiers

πŸ“Š Mapping Types

The mapping type sets what kind of source the mappings made against an RDM refer to:

TypeSources on mappings are
VENDORVendors
PRODUCTProducts
FEATUREProduct features
CONTROLControls
INTELIntelligence
GENERICGeneric information

πŸ“Œ All of the RDMs currently in production use the PRODUCT type.


🏷️ Relationship Traits

Traits define the core relationship types that can exist between an entity and a framework node.

Each trait includes:

  • A machine-readable value (e.g., PROTECT), up to 20 characters matching ^[A-Z][A-Z0-9-]{1,18}[A-Z]$
  • A human-readable label (e.g., Protect), up to 30 characters
  • Criteria that define when the trait should be applied, up to 2,000 characters

⚑ Traits are the primary way to express relationships. An entity must have at least one trait relationship to a framework node.


πŸ”„ Relationship Modifiers

Modifiers allow for more nuanced expression of relationships by adding qualifiers.

Each modifier includes:

  • A machine-readable key (e.g., EFFECTIVENESS), up to 20 characters matching ^[A-Z][A-Z0-9-]{1,18}[A-Z]$
  • A human-readable label (e.g., Effectiveness), up to 30 characters
  • A description of the modifier's purpose, up to 2,000 characters
  • Whether the modifier is required
  • A set of 1-20 possible values (traits) that can be assigned

πŸš€ Example: Cybersecurity Capability RDM

A common RDM for security tooling might define:

Relationship Traits:

  • IDENTIFY: Tool helps identify assets, risks, or threats
  • PROTECT: Tool provides protective capabilities
  • DETECT: Tool detects security events or anomalies
  • RESPOND: Tool assists in security incident response
  • RECOVER: Tool supports recovery from security incidents

Modifiers:

  • EFFECTIVENESS (Required): How effectively the tool performs the primary function
    • Values: SIGNIFICANT, PARTIAL, MINIMAL

πŸ“Œ This modifier is illustrative. None of the RDMs currently in production define any modifiers.

πŸ“„ Example: MITRE ATT&CK Enterprise RDM

The production RDM for MITRE ATT&CK Enterprise lives at prod-files/mitre/attack/enterprise/rdm_mitre-attck-enterprise.json in data-fdm-converter. The sample below shortens the criteria and shows two of its five traits.


πŸ’‘ Practical Applications

RDMs enable flexible relationship modeling:

  • Versioning: Multiple RDMs of a given type can exist, enabling versioning or iterations
  • Custom Mappings: Additional RDMs can be defined for existing frameworks
  • AI Integration: The trait criteria (and any modifiers) of an RDM are passed to ESPROFILER's AI-powered mapping engine, which uses them to choose a relationship for each node. AI mapping currently supports PRODUCT and FEATURE RDMs
  • Standardization: Creates consistent vocabulary across different mapping exercises

πŸ“Œ In practice, a framework has a default RDM defined against a type (e.g., PRODUCT). The first RDM created for a type becomes the default, and a different RDM of that type can be made the default later. Products in a capability exchange would then be mapped against this RDM and framework.

Copyright Β© 2026