Platform Framework Matrix V2
The framework matrix should present 4 main statistical information:
- Coverage (and a break down by relationship)
- Product Cost
- Product Count
- Priority (with regards to threats)
The matrix was designed as an instance of the CentralFramework object utilized by MFS and CPS. The CentralFramework also has a list of nodes which are an object of type CentralFrameworkNode.
PlatformFrameworkMatrix or CentralFrameworkMatrix class, and CPS has no matrix endpoint. The view was built in platform-api as the Framework Collider: FrameworkCollider, FrameworkColliderNode and FrameworkColliderAnalytics in com.esprofiler.platform.capability.collider, served from GET /api/v1/collider/{frameworkRoot}. In the platform UI it is the Frameworks page under Security stack (route /portfolio/frameworks/:framework), with an Overview tab (products, vendors, annualized cost and coverage) and a Framework tab (the node canvas with coverage, priority and product filters). The sections below keep the original design. JSON samples and the field-by-field build notes are in the dropdowns.CentralFramework and CentralFrameworkNode do exist, in MFS (com.esprofiler.mfs.centralframeworks.fdm.dto.output) and in platform-api (com.esprofiler.platform.frameworks.fdm.dto.output). Both extend the shared Framework and FrameworkNode models. A CentralFramework serializes with its id, root, name, version, type, privacy and nodes. Null fields are omitted.
{
"id": "uuid",
"root": "NIST-CSF",
"name": "NIST CSF 2.0 Framework",
"version": "V2.0.0",
"type": {
"value": "COMPLIANCE",
"label": "Compliance",
"description": "Frameworks that define regulatory requirements and compliance standards for organizations"
},
"privacy": {
"value": "PUBLIC",
"label": "Public",
"description": "The model is accessible by anyone"
},
"nodes": [
{
"id": "uuid",
"node": "GV",
"name": "GOV",
"state": {
"type": "ACTIVE"
},
"leaf": false,
"path": "path string",
"nodes": []
}
]
}
Notes on the example:
- A node has no
selffield. Itsselfhash is held inside itspathlocation. stateis an object (ACTIVE,DEPRECATEDorREPLACED), not null.pathis serialized as a string. Its location values (self,parent,family,base,chainanddepth) are hashes used internally.nodesis a set of nodes, serialized as an array at the top level and on each branch node for its children.- The full MFS output also includes
organisationandversions(and, when not compact,description,licence,url,authorsandmeta).
The version defaults to the latest version of the framework when the root is given with no version or as root~ (for example NIST-CSF~). This is resolved in CentralFrameworkModalService.frameworkModelGet using PathParser.ParsedPath.isLatestVersion(). The collider controller appends ~ to a root that has none.
This object however was to be overloaded to include the mapping information when retrieved as a matrix. Hence the node object would have a list field called mappings which can be empty to indicate there is no mapping on it and an rdm object indicating what RDM the matrix is loaded with. In addition to this field would be the statistical information mentioned above. The dropdown has the original PlatformFrameworkMatrix sample (some fields from CentralFramework and CentralFrameworkNode are omitted). It is the original design. The shape the API returns is in the build notes below it.
{
"id": "uuid",
"root": "NIST-CSF",
"name": "framework name",
"version": "V2.0.0",
"rdm": {
"id": "uuid",
"type": "PRODUCT",
"name": "NIST CSF 2.0 Function-Based Capability Model",
"relations": [
{
"value": "PROTECT",
"label": "Protect",
"description": "description"
}
],
"modifiers": {}
},
"analytics": {
"priority": 0.56,
"sourceCount": 20,
"coverage": {
"PROTECT": {
"count": 23,
"coverage": 0.84,
"sources": ["uuid1", "uuid2"],
"cost": {
"amount": 1331207.280225,
"currency": "GBP",
"formatted": "£1,331,207.28"
}
},
"DETECT": {
"count": 20,
"coverage": 0.16,
"cost": {
"amount": 1331207.280225,
"currency": "GBP",
"formatted": "£1,331,207.28"
}
}
}
},
"nodes": [
{
"id": "uuid",
"node": "GV",
"state": null,
"path": {},
"analytics": {
"priority": 0.56,
"sourceCount": 20,
"coverage": {
"PROTECT": {
"count": 23,
"coverage": 0.84,
"sources": ["uuid1", "uuid2"],
"cost": {
"amount": 1331207.280225,
"currency": "GBP",
"formatted": "£1,331,207.28"
}
},
"DETECT": {
"count": 20,
"coverage": 0.16,
"cost": {
"amount": 1331207.280225,
"currency": "GBP",
"formatted": "£1,331,207.28"
}
}
}
},
"nodes": {}
}
]
}
In this example:
rdmholds the RDM information the matrix is loaded with.pathholds the node's Path object.analytics.priorityis a percentage value of the aggregated priority on nodes.analytics.sourceCountis the count of unique sources.- Under
analytics.coverage, each relation has:count: the integer count of mappings with this relationcoverage: a percentage value of the aggregated coverage on nodessources: the ids of the sources mapped with this relationcost: the cost information
As built, analytics.coverage is a single float, the breakdown by relationship is relationshipDistribution, cost is one total, and nodes have no mappings list. FrameworkColliderService in platform-api builds the response itself. Field names, the original CPS request, and the endpoint steps are in the dropdowns.
As built, FrameworkCollider (extends Framework) and FrameworkColliderNode (extends FrameworkNode) differ from the design above:
- The top level has
id,root,version,name,type,privacy,description,rdm,analytics,index,unmappedProductCount,totalProductCountandnodes. rdmis the sharedRelationDefinitionmodel:rdm(the RDM key, notid),type,name,description,relationsandmodifiers. Each relation is{value, label, criteria}, so the field iscriteria, notdescription.modifiersis an array, not an object, and each modifier is{key, label, description, required, values}wherevaluesis a list of{value, label, criteria}. The RDM is the framework's defaultPRODUCTRDM.analyticsis a flat object. Its fields aretotalLeafNodes,totalLeafCovered,mappingCount,vendorCount,alignment,priority,coverage,confidence, thereality*fields (realityCoverage,realityAchieved,realityStatus,realityDistribution,realityFullyAssessed,realityClaimsAssessedandrealityClaimsTotal),cost,actionability,priorityBreakdown,relationshipDistributionandrelatedProducts. There is nosourceCount; the nearest arevendorCountand the product and vendorindex.analytics.coverageis a single float (0 to 1), not a map by relation. The break down by relationship isrelationshipDistribution, a map of relation value to mapping count.analytics.costis one total{amount, currency, formatted}value. A per-relationcostandsourcesare not returned.relatedProductslists, per product,product(id),mappingCount,confidence,costandrelationshipDistribution.- Nodes have
id,node,name,state(an object),leaf,path(a string),analyticsand, on branch nodes,nodesas an array of child nodes (not{}). Leaf nodes have nonodes. The top-levelnodesis also an array. - There is no
mappingslist on nodes. Mappings are aggregated into each node'sanalytics.
The implementation would make a request to CPS with:
root~version: The framework root and versionrdm: The RDMuuidto request for.product.id: a set of product ids to fetch mappings. If empty, send the full platform inventory (with filters)
CPS would build the CentralFrameworkMatrix which will have the mappings, coverage & count information.
Platform would receive this object and build up a PlatformFrameworkMatrix composing it with commercial and threat data, which will be used to populate the cost and priority information.
Not implemented as designed. There is no CentralFrameworkMatrix in CPS, and platform-api does not call CPS for the collider. FrameworkColliderService in platform-api does the work itself:
GET /api/v1/collider/{frameworkRoot}takes the framework root (aroot~versionstring is accepted and~means latest), an optionalsearch, the filters inFrameworkColliderQuery(relationship,operationalStatus,priorityGroupId,minPriority,maxPriorityand the FDM node query) and the product filters inLicensedProductQuery(includingscenarioId). There are nordmorproduct.idparameters.- It loads the framework tree, resolves the framework's default
PRODUCTRDM (CentralRelationService.rdmDefaultGet) and selects the products matching the filters (the full licensed inventory when there is no filter). - Mappings, mapping counts and relationship counts per leaf node are read from the capability collection mappings with
CapabilityCollectionMappingRepo, and per-product cost withProductLicensedRepo.listProductMappingCost. - Priority comes from
PriorityNodeStatsRepo(threat priority statistics, optionally limited to priority groups). Cost is accumulated byProductCostVisitorand coverage byCoverageVisitor(a leaf is covered or not, and a branch is the average of its children).
Related endpoints are GET /api/v1/collider/{frameworkRoot}/summary and GET /api/v1/collider/{frameworkRoot}/export (a MITRE Navigator export, for the MITRE ATT&CK frameworks only).
Subsequently, more product information such as utilization from assertion and deep research can be paired with this to show even greater insights. The collider already includes the reality* fields (derived from assessed mapping claims) and product utilization and sentiment in the index.
Relationship Mapping Model (RMM)
The Relation Mapping Model (RMM) is the concrete implementation that creates actual relationships between sources and framework nodes. It leverages the vocabulary defined in a Relationship Definition Model (RDM) to establish specific, contextual connections.
Glossary
This is a collection of acronyms and terms that will come up regularly in your day-to-day life at ESProfiler.

