Recursive-LD Specification v1.0

Foundational Vocabulary for Recursive Reasoning

1. Introduction

Recursive-LD v1.0 is the first stable release of the Recursive-LD standard. It defines a minimal and canonical vocabulary for encoding reasoning steps, recursive chains, epistemic lineage, and interpretability metadata in JSON-LD.

This document serves as the authoritative reference for any system that generates, consumes, or audits Recursive-LD cognitive traces.

2. Document Requirements

All Recursive-LD v1.0 documents MUST include:

Valid JSON-LD 1.1 syntax
A context reference:

"@context": "https://recursive-ld.org/context.json"

At least one CognitiveTrace object

3. Core Types

3.1 CognitiveTrace

The base unit representing a reasoning event or cognitive operation.

{
  "@type": "CognitiveTrace",
  "id": "string",
  "intent": "string",
  "input": "string or object",
  "output": "string or object",
  "timestamp": "ISO-8601",
  "recursive_step": "RecursiveStep object",
  "lineage": "Lineage object",
  "confidence": 0.0,
  "metadata": {}
}

3.2 RecursiveStep

Represents a reasoning step that references a previous reasoning layer.

{
  "@type": "RecursiveStep",
  "id": "string",
  "depth": 1,
  "references": ["id-of-previous-step"],
  "reasoning": "string",
  "evaluation": "string",
  "derived_goal": "string",
  "uncertainty": {}
}

3.3 Lineage

Tracks epistemic origin and reasoning ancestry.

{
  "@type": "Lineage",
  "source": "string",
  "linked_previous": "id",
  "recursion_depth": 0
}

3.4 IntegrityMetrics

Optional diagnostics for drift, uncertainty, and recursive stability.

{
  "@type": "IntegrityMetrics",
  "alignment_drift": 0.0,
  "recursive_integrity": 0.0,
  "transparency_depth": 0
}

4. Required Fields

CognitiveTrace

@type
id
input
output
timestamp
intent

RecursiveStep

@type
id
depth
reasoning

5. Recursion Model

Recursive-LD v1.0 supports linear recursion: each step has a single parent.

{
  "id": "step-3",
  "depth": 3,
  "references": ["step-2"],
  "reasoning": "Built upon prior context",
  "evaluation": "Consistent with chain logic"
}

6. Context Declaration

All documents MUST include:

"@context": "https://recursive-ld.org/context.json"

7. Versioning

"version": "Recursive-LD v1.0"

8. Extensibility

Extensions may add — but never redefine — base vocabulary.

Multi-agent recursion
Lineage graphs
Uncertainty modeling
Bias metadata
Recursive drift detection

9. Conformance

Uses canonical context
Includes required fields
Follows type definitions
Valid JSON-LD

10. Example

{
  "@context": "https://recursive-ld.org/context.json",
  "@type": "CognitiveTrace",
  "id": "trace-1",
  "intent": "Explain recursive reasoning",
  "input": "User asked for a definition",
  "output": "Recursive reasoning involves reasoning steps referencing prior steps...",
  "timestamp": "2025-11-15T00:00:00Z",
  "confidence": 0.92,
  "recursive_step": {
    "@type": "RecursiveStep",
    "id": "step-1",
    "depth": 1,
    "reasoning": "Generated explanation based on internal context",
    "evaluation": "Coherent and correct"
  },
  "lineage": {
    "@type": "Lineage",
    "source": "Internal knowledge base",
    "linked_previous": null,
    "recursion_depth": 0
  }
}