Skip to main content

📘 DPFI SYNCTX Inference Rules (Formal Logic)

1. Purpose

Define how the HDIP platform (DPFI) infers, constructs, and validates SYNCTX from declarative product artifacts.

2. Inputs to Inference Engine

Let’s formalize inputs:

IntentRecord (IR)
SemanticDescriptor (SD / DPROD draft)
PolicyIntent (PI)
BusinessAssertions (BA)
UsageContract (UC)
ProductDependencies (PD)

3. Output

SYNCTX = SyntheticContext Graph

4. Core Inference Model

We define SYNCTX as:

SYNCTX = f(IR, SD, PI, BA, UC, PD)

Where:

  • f = deterministic inference function
  • output = graph (not document)

5. Inference Phases

DPFI executes SYNCTX inference in 5 ordered phases:

🔹 Phase 1 — Synthetic Requirement Detection​

Rule 1.1 — Synthetic Trigger (Usage-Based)​

IF UC.purpose ∈ {model_training, external_sharing, testing, simulation}
THEN synthetic_required = TRUE

Rule 1.2 — Multi-Product Dependency Trigger​

IF PD.count > 1
AND synthetic_required = TRUE
THEN context_required = TRUE

Rule 1.3 — Explicit Constraint Trigger​

IF PI contains (privacy_constraint OR no_pii OR masking_required)
THEN synthetic_required = TRUE

Result:​

IF synthetic_required = TRUE AND context_required = TRUE
THEN proceed to SYNCTX inference
ELSE skip SYNCTX

🔹 Phase 2 — Member Resolution

Rule 2.1 — Direct Membership​

SYNCTX.members = PD.products

Rule 2.2 — Transitive Dependency Expansion​

FOR each product p in PD:
    INCLUDE all upstream dependencies(p)

Rule 2.3 — Deduplication​

SYNCTX.members = unique(products)

🔹 Phase 3 — Relationship Inference (Critical)

Rule 3.1 — Semantic Relationship Extraction​

FOR each pair (p1, p2) in SYNCTX.members:
    IF SD defines relationship(p1, p2)
    THEN add Relationship(p1 → p2)

Rule 3.2 — Ontology-Based Inference​

IF ontology contains relationship(p1.entity, p2.entity)
THEN infer relationship

Rule 3.3 — Key Alignment Rule​

IF attribute(p1, keyX) matches attribute(p2, keyX)
THEN relationshipType = foreign_key

Rule 3.4 — Referential Integrity Requirement​

IF relationship exists
THEN mark:
  preserve_referential_integrity = TRUE

🔹 Phase 4 — Generation Strategy Inference

Rule 4.1 — Mode Selection​

IF relationships.count > 0
THEN generationMode = coordinated
ELSE generationMode = independent

Rule 4.2 — Fidelity Determination​

IF UC.purpose = model_training
THEN fidelity = high

IF UC.purpose = testing
THEN fidelity = medium

IF UC.purpose = sandbox
THEN fidelity = low

Rule 4.3 — Preservation Rules​

IF generationMode = coordinated
THEN preserve:
  - referential_integrity
  - statistical_distribution

🔹 Phase 5 — Constraint Compilation

Rule 5.1 — Policy Translation​

FOR each policy in PI:
    translate → constraint

Rule 5.2 — Privacy Constraint​

IF PI contains no_pii
THEN constraint:
  remove_pii = TRUE

Rule 5.3 — Compliance Constraint​

IF regulatory_requirement exists
THEN enforce constraint(region, retention, masking)

🔹 Phase 6 — Output Mapping

Rule 6.1 — Synthetic Product Creation​

FOR each product p in SYNCTX.members:
    CREATE synthetic_product sp
    sp.basedOn = p

Rule 6.2 — Context Binding​

FOR each synthetic_product sp:
    sp.synctxRef = SYNCTX.id

🔹 Phase 7 — Validation Rules

Rule 7.1 — Minimum Membership​

IF SYNCTX.members < 1
THEN INVALID

Rule 7.2 — Joinability Check​

IF relationships.count = 0 AND members.count > 1
THEN WARN: non-joinable context

Rule 7.3 — Consistency Check​

IF conflicting relationships exist
THEN FAIL

6. Formal Logical Representation (Compact)

Predicate Form​

SyntheticRequired(IR, PI, UC)
ContextRequired(PD, SyntheticRequired)
Members = ResolveMembers(PD)
Relationships = InferRelationships(SD, Ontology)
Generation = InferGeneration(UC, Relationships)
Constraints = CompileConstraints(PI)
Outputs = GenerateSyntheticProducts(Members)

SYNCTX = {
  Members,
  Relationships,
  Generation,
  Constraints,
  Outputs
}

7. Graph Construction Mapping

Members → synctx:hasMemberProduct
Relationships → synctx:hasRelationship
Outputs → synctx:produces
Constraints → synctx:hasConstraint

8. Determinism Principle

Given same inputs, SYNCTX MUST be identical

f(inputs₁) = f(inputs₂) → SYNCTX₁ = SYNCTX₂

9. Observability

DPFI MUST log:

  • why SYNCTX was triggered
  • which relationships were inferred
  • which policies created constraints
  • why generation mode was chosen