Intertween.*

Relationships and the register

What connects, who says so and on what grounds? The relational metadata register describes Knowledge Objects and their relationships. A connection needs a traceable asserting party, evidence and context.

A relationship is also an assertion

Relational means that connections carry meaning and context. It does not prescribe a database product or storage technology. A relationship such as “derived from” needs to be inspectable: who asserted it, by what method, when and on what evidence?

A source may explicitly cite another. A person can make an assessment, and an AI can suggest a connection. These ways of knowing a relationship need to remain distinct. The bill example shows records with evidence in one source. An intertween goes a step further: a new claim or question that depends on several independent sources receives its own identity, evidence and review status.

See how the question between sources gets its own place ↗

Finding the file is only the beginning

A guideline may be easy to find but difficult to interpret. Which legal provision supports it? Which version applies? Who is responsible? Is there a challenge or a later assessment? The register connects these descriptions and references to identifiable Knowledge Objects.

In the proposed architecture, each node maintains its permitted descriptions and references. The register is a function within the node. It provides a shared way to describe context across systems. Documents, measurement series and other source material need not move into a shared content store.

PROPOSED TEACHING VIEW · FICTIONAL VALUES

A relationship to examine

From → to
EX-METOD-01, version 1 → EX-OBS-01, version 1
Assertion
The method is derived from the observation
Asserting role and method
Reviewer role A · manual assessment
Evidence
Reference to section 2 of the fictional source
Recording and status
14 January 2026 · proposed, not yet reviewed
Conditions
Illustration only. No real rights or permissions are enforced.

The view explains the questions we need to ask. It is not a frozen schema or a deployed register service.

Descriptions that can be followed

What the register describes
PartQuestions it helps us ask
Knowledge ObjectsWhich document, assertion, dataset or method is described? Where are its source and representations?
Actors and contextWho contributed, who maintains the source and in what context was it created?
RelationshipsWhat references, derives from, challenges or supersedes something else? Who asserts the connection?
Conditions and responsibilityWhich purposes, rights and mandates need to be considered?
Current statesWhich version is referenced? Is the claim challenged or superseded? What remains unknown?

The register describes people, organisations, things and processes through references and Knowledge Objects. A description remains distinct from the person, organisation or phenomenon it concerns.

From two guidelines to a reusable method

Illustrative scenario with fictional municipalities.

  1. 01

    Two guidelines

    Municipalities A and B each hold a guideline. The register describes sources, versions, responsibility and their references to the same legal provision. Source material remains with each municipality.

  2. 02

    A difference raises a question

    The guidelines specify different distance thresholds. An intertween describes the question: what might explain the difference? References identify the supporting material.

  3. 03

    A reviewer adds context

    A reviewer points out that local circumstances are missing from the comparison. The contribution links to the question with its source and explicit status. In the proposed model, others could see the contribution and evidence of the reviewer’s mandate without receiving their name. The difference does not establish that either guideline is wrong.

  4. 04

    A candidate skill

    A method is proposed: check the shared source, local context and missing information before judging differences. It gets its own Knowledge Object, linked to the question and the reviewer’s contribution.

  5. 05

    Evaluation and further learning

    Evaluations, adoption decisions and later outcomes have separate references. The original question remains. New evidence can prompt reassessment or further candidate methods.

The register describes the current state of these connections. The event history records how their descriptions changed. This is a branching network, not a mandatory sequence.

Follow the challenge in the scripted example ↗

Follow terminology without losing history

FICTIONAL TEACHING EXAMPLE

Different terms can lead to a shared question

The municipality’s source and the transport organisation’s source are different Knowledge Objects with their own references. A possible connection needs review even when their descriptions sound similar.

Municipal source · SK-24

Individual school transport assessment

Version 1 · the term is unchanged

Transport organisation source · FT-71

Exception assessment

Version 1

Two sources, two references and a connection to investigate. Similar terms or identifiers do not establish equivalence.

Terms and references are fictional. The change is prepared and does not connect to a real register.

Three distinct functions

Proposed architecture. These are roles, not three deployed services.

The register, history and discovery
FunctionMain question
Relational metadata registerWhat exists, how does it connect and what state is described now?
Relational TraceWhat was recorded as happening, by whom and how did the records change?
Federated indexWhat permitted information can be discovered across nodes, and where are its sources?

A search result is not permission to read a source. The register’s current view does not replace history, and the index does not establish truth. Relational Trace is the profile for a separate attributable history with its own integrity requirements.

Shared context, different questions

For people

Find material, understand its relevance, inspect a connection, see what is unknown and identify who handles a request or challenge.

For machines

Use structured identifiers and relationships to select permitted context, follow dependencies and reference the relevant version. Suggested new connections still need evidence and review.

If a source is corrected, dependencies can help identify assessments and skills for reassessment. A connection indicates what merits investigation; it does not prove every downstream conclusion wrong.

Cooperation with local responsibility and assessed rights

Finding an object does not grant access to its source. Reading does not automatically grant permission for reuse or training. Metadata can itself be sensitive. The existence of an object, a contributor’s identity or the relationship between two objects may need protection. A recipient’s view needs assessment against purpose, authority and conditions. Discovery, access, reuse and training are distinct questions.

The register is a descriptive function in the architecture. No ready-to-use node service is available here. The examples and validator do not establish full exchange between nodes or enforcement of usage rights. Metadata can describe conditions, but descriptions do not replace enforcement. Rules must be enforced where information is disclosed and activities are executed.

A public view, a restricted view for an authorised recipient and a locally held record may therefore contain different details about the same object. The proposal also requires review of relationships: date, place and role can jointly reveal an identity. Local sovereignty means responsible parties assess disclosure and purpose; it does not give control over every copy already shared.

Conditions may need review when a mandate ends or a purpose changes. A review date should raise a decision for the responsible custodian, not automatically delete material. Nodes would need to acknowledge changes or explain why they have not applied them. This workflow remains to be built and tested.

Proposed exchange between two nodes
  1. Node A stewards the source and assesses which descriptions may be disclosed.
  2. A permitted metadata view makes relevant material discoverable from node B. A relationship may itself need to be withheld.
  3. Node B requests access for a stated purpose. A search result is not an access decision.
  4. The responsible source party assesses the request. Permitted content is disclosed; a denied request must not reveal protected metadata.
  5. Changed conditions need handling and acknowledgement. This does not give control over every copy already shared.

Architecture outline. This is not an executable exchange or a demonstration of working rights enforcement.