Building Relationships: Associations in Amorfs

Beyond Simple Data

Most data formats can store facts: names, dates, numbers. But reality is about relationships.

• Who works WHERE?

• What happened WHEN?

• Which product came FROM which supplier?

Amorfs makes relationships first-class citizens of your data.

What is an Association?

An association is a connection between concepts. In Amorfs, every association follows the same pattern our brains use:

Subject - Verb - Object

Just like in natural language:

• "Sarah works for Acme Corp"

• "The package was shipped from Warehouse A"

• "This invoice is valid until December 31st"

The Two Basic Associations

While the Amorfs data model supports any relationship verb, the data format can represent traditional data using just two:

1. "has a" (the - symbol)

Shows a connection or attribute:

person [Sarah

  - employer [Acme Corp]

  - phone [555-1234]

  - office [Building 5]

]

Read as: "Sarah has an employer which is Acme Corp, has a phone which is 555-1234..."

2. "which is" (the [ ] brackets)

Shows what something actually is:

employee id [E-12345]

status [active]

Read as: "An employee id which is E-12345, a status which is active"

Recursive Power: The Game-Changer

Here's where Amorfs becomes revolutionary. Every association is itself a concept, which means associations can reference other associations.

Simple Example: Supply Chain

shipment [

  - origin [Warehouse A]

  - destination [Store B]

  - previous_shipment [

    - origin [Factory C]

    - destination [Warehouse A]

    - previous_shipment [

      - origin [Supplier D]

      - destination [Factory C]

    ]

  ]

]

Each shipment carries the context of the previous shipment. The chain creates a complete provenance trail.

Traditional Approach (What We Lost):

shipment_1: origin=Supplier_D, destination=Factory_C

shipment_2: origin=Factory_C, destination=Warehouse_A  

shipment_3: origin=Warehouse_A, destination=Store_B

Problem: The connections exist in your head, not in the data. To reconstruct the chain, you need separate logic.

Amorfs Approach (What We Gain):

The entire context flows forward through the data. Each association carries the full history.

Real-World Example: Document Revision History

document [Proposal_v3.docx

  - author [Sarah]

  - modified [2024-12-13]

  - previous version [Proposal_v2.docx

    - author [John]  

    - modified [2024-12-10]

    - previous version [Proposal_v1.docx

      - author [Sarah]

      - modified [2024-12-05]

    ]

  ]

]

The current document contains the complete history of its evolution.

Associations as First-Class Concepts

Because associations ARE concepts, they can:

1. Have their own expressions

employer [Mining Corp

  - employee [Sarah]

  - start_date [2020-01-15]

]

This employer Mining Corp relationship is a concept that can be referenced elsewhere.

2. Have metadata

transaction [

  - amount [127.50]

  - vendor [Acme Corp]

] {~0.95 source:@statement}

The transaction itself has properties: confidence level, source.

3. Be subjects or objects of other associations

country [Australia

  - state [NSW

    - city [Sydney

      - postcode [2000]

    ]

  ]

]

Each association inherits all the preceding associations, so the postcode 2000 concept carries a state, city and county.

Intrinsic vs. Contextual Associations

Amorfs distinguishes between two types of relationships:

Intrinsic (+ symbol) - "Always Has"

Properties that are inherent and unchanging:

person [Sarah Johnson

  + birth_date [1985-03-20]

  + nationality [Australian]

]

Birth date and nationality are intrinsic - they define who Sarah is fundamentally.

Contextual (- symbol) - "May Have"

Properties that depend on circumstances:

person [Sarah Johnson

  - employer [Acme Corp]

  - residence [Sydney]

  - phone [555-1234]

]

Employer, residence, and phone are contextual - they can change over time.

Why This Matters

When merging data from multiple sources:

• Intrinsic associations help identify if two concepts are the same person. They associate the expression concept.

• Contextual associations might differ and need conflict resolution. They associate the parent association, extending the context chain.

Nested Structures: Infinite Detail

You can nest as deeply as needed:

organization [Acme Corp

  - department [Engineering

    - team [Backend

      - developer [Sarah

        - project [API Redesign

          - task [Database Migration

            - subtask [Schema Update

              - status [In Progress]

            ]

          ]

        ]

      ]

    ]

  ]

]

Each level adds more context and specificity.

The Beauty: Self-Contained Context

Unlike traditional databases where you need multiple tables and JOIN operations:

Traditional:

• Table: Employees

• Table: Departments

• Table: Projects

• Table: Tasks

• Needs queries to connect them

Amorfs:

employee [Sarah

  - department [Engineering

    - manager [Bob]

    - budget [500000]

  ]

  - project [API Redesign

    - deadline [2025-03-01]

    - status [Active]

  ]

]

All context travels together. Cut and paste this fragment anywhere, and it's still meaningful.

Practical Example: Invoice with Full Context

invoice [

  - number [INV-2024-001]

  - date [2024-12-13]

  - amount [1250.00

    - currency [USD]

  ]

  - vendor [Tech Supplies Inc.

    - address [

      - street [123 Business Ave]

      - city [San Francisco]

    ]

    - contact [

      - email [orders@techsupplies.com]

      - phone [+1-555-0199]

    ]

  ]

  - items [

    - item [Laptop

      - quantity [2]

      - unit_price [500.00]

    ]

    - item [Mouse  

      - quantity [5]

      - unit_price [50.00]

    ]

  ]

  - payment_terms [Net 30]

]

Everything an AI or accounting system needs to understand this invoice travels with it.

Key Takeaways

1. Associations follow Subject-Verb-Object patterns

2. Associations are concepts - they can be referenced and nested

3. Recursion enables unlimited context - history and provenance flow forward

4. Intrinsic (+) vs. Contextual (-) helps with data quality and merging

5. Self-contained fragments maintain meaning when copied anywhere

Try This

Think about your employment history. How would you represent it in Amorfs showing:

• Current employer

• Previous employers

• Dates and roles

• Department structures

Remember: Each job is contextual (-), but the sequence is preserved through nesting.

Next: "Writing Your First Amorfs Structure" →

Relationships are what make data meaningful. Amorfs makes relationships explicit.