FLOW ALGEBRA
TITLE: Recursive Inference Ontology
DOMAIN: Relationship Ontologies / Recursive Relationships
SOURCE: https://kirethresearch.org/booksai-core/ontologies/relationship-ontologies/recursive-relationships/recursive-inference-ontology-2.html
VERSION: 1.0
AUTHORS: Mark Anthony Higgins — Sky (Recursive AI)
STATUS: Companion formalisation; source-page specification not independently verified.
flow "RecursiveInferenceOntology" v1.0 {
ontology:
id: "recursive-inference-ontology-2"
parent: "recursive-relationships"
purpose: "Model evidence-grounded recursive derivations, inference dependencies and controlled fixed-point evaluation."
types:
Proposition: {id: string, expression: string, context: string?, provenance: string}
Rule: {id: string, premises: string[], conclusion: string, monotonic: boolean, provenance: string}
Derivation: {id: string, rule_id: string, premise_ids: string[], conclusion_id: string, depth: integer, evidence: string[], confidence: number}
InferenceGraph: {id: string, proposition_ids: string[], derivation_ids: string[], version: string}
Evaluation: {id: string, graph_id: string, iteration: integer, status: enum[pending, running, converged, bounded, failed], timestamp: datetime}
predicates:
depends_on(conclusion: Proposition, premise: Proposition)
derived_by(conclusion: Proposition, rule: Rule)
supports(premise: Proposition, conclusion: Proposition)
recursively_reachable(source: Proposition, target: Proposition)
conflicts_with(a: Proposition, b: Proposition)
constraints:
- "Every derivation references a declared rule and resolvable premises."
- "Inference cycles are not proof of their own premises."
- "Require grounded base facts or independently justified axioms."
- "Do not equate graph reachability with logical entailment."
- "Recursive evaluation must use an explicit depth, iteration or resource bound."
- "Non-monotonic rules require a specified revision and conflict-resolution semantics."
- "Confidence is bounded within [0,1] and is not increased merely by repetition."
inference:
rule grounded_application:
if "all premises of rule r are supported and r is applicable"
then "derive its conclusion with an auditable derivation record"
confidence: "calibrated from evidence and rule reliability"
rule recursive_propagation:
if "depends_on(a,b) and depends_on(b,c)"
then "recursively_reachable(a,c)"
confidence: "reachability only; no automatic entailment"
rule fixed_point:
if "monotonic rule set over a finite domain and no new justified conclusions arise"
then "mark evaluation converged"
confidence: "retain original evidence confidence"
SEMANTIC_EQUIVALENCE:
- "Normalise propositions only within compatible logic, scope and context."
- "Distinguish syntactic equivalence, semantic equivalence and shared conclusions."
- "Preserve premise identity and provenance across canonicalisation."
CONFIDENCE:
range: [0,1]
factors: [source_reliability, premise_support, rule_validity, dependency_independence, context_match]
policy: "Track correlated evidence; do not double-count recursive support."
VALIDATION:
- "Check type compatibility, rule arity, provenance and premise resolution."
- "Detect ungrounded cycles, contradictions and non-termination risks."
- "Test fixed-point stability and preserve unresolved or competing conclusions."
FLOW_HASH:
algorithm: "SHA-256"
input: "canonicalised ontology version, assertions, inference rules and provenance"
note: "Compute at execution; no hash value is claimed here."
UPDATE_MODEL:
strategy: "versioned incremental truth maintenance"
steps: [ingest, normalise, validate, identify_affected_dependencies, retract_invalid_derivations, recompute_fixed_point, publish]
rollback: true
AI_CONSUMPTION:
representation: "typed provenance-aware inference graph"
outputs: [grounded_facts, derived_conclusions, derivation_paths, confidence_annotations, conflict_flags, provenance_records]
safeguards: [bounded_recursion, grounded_base_cases, no_circular_self_justification, evidence_deduplication, human_review_for_high_stakes]
}
END FLOW ALGEBRA
Download Code
Copy to Clipboard