GCMD_Keyword_Classifier_MVP / schemas /dataset_coverage_resolution.schema.json
igerasimov's picture
Deploy deterministic dataset README classifier
4af38c8
Raw History Blame Contribute Delete
23.3 kB
{
"$defs": {
"AliasSource": {
"description": "Approved provenance for a product identity alias.",
"enum": [
"validated_cmr_identity",
"deterministic_derivation",
"deterministic_normalization",
"readme_declaration"
],
"title": "AliasSource",
"type": "string"
},
"ConfidenceSignal": {
"additionalProperties": false,
"description": "Explicitly uncalibrated confidence signal.",
"properties": {
"value": {
"maximum": 1.0,
"minimum": 0.0,
"title": "Value",
"type": "number"
},
"calibrated": {
"const": false,
"default": false,
"title": "Calibrated",
"type": "boolean"
}
},
"required": [
"value"
],
"title": "ConfidenceSignal",
"type": "object"
},
"CoverageMode": {
"description": "Supported README document coverage modes.",
"enum": [
"single_product",
"explicit_multi_product",
"product_family",
"vague_product_scope",
"undetermined"
],
"title": "CoverageMode",
"type": "string"
},
"DatasetIdentity": {
"additionalProperties": false,
"description": "Authoritative CMR collection identity.",
"properties": {
"concept_id": {
"minLength": 1,
"title": "Concept Id",
"type": "string"
},
"native_id": {
"minLength": 1,
"title": "Native Id",
"type": "string"
},
"short_name": {
"minLength": 1,
"title": "Short Name",
"type": "string"
},
"version": {
"minLength": 1,
"title": "Version",
"type": "string"
},
"cmr_revision_id": {
"minimum": 1,
"title": "Cmr Revision Id",
"type": "integer"
},
"legacy_entry_id": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"title": "Legacy Entry Id"
}
},
"required": [
"concept_id",
"native_id",
"short_name",
"version",
"cmr_revision_id"
],
"title": "DatasetIdentity",
"type": "object"
},
"DatasetStage": {
"description": "Dataset workflow stages used by diagnostics.",
"enum": [
"source",
"readme_discovery",
"readme_retrieval",
"extraction",
"product_resolution",
"evidence_selection",
"classification",
"validation",
"blind_persistence",
"comparison"
],
"title": "DatasetStage",
"type": "string"
},
"DatasetWarning": {
"additionalProperties": false,
"description": "Non-fatal dataset workflow diagnostic.",
"properties": {
"code": {
"minLength": 1,
"title": "Code",
"type": "string"
},
"message": {
"minLength": 1,
"title": "Message",
"type": "string"
},
"stage": {
"anyOf": [
{
"$ref": "#/$defs/DatasetStage"
},
{
"type": "null"
}
],
"default": null
},
"details": {
"anyOf": [
{
"additionalProperties": true,
"type": "object"
},
{
"type": "null"
}
],
"default": null,
"title": "Details"
}
},
"required": [
"code",
"message"
],
"title": "DatasetWarning",
"type": "object"
},
"DeclarationKind": {
"description": "Deterministically recognized declaration structure.",
"enum": [
"identity",
"enumerated_products",
"product_family",
"document_wide",
"other_product_only"
],
"title": "DeclarationKind",
"type": "string"
},
"DeclarationLocality": {
"additionalProperties": false,
"description": "Deterministic forward-only applicability range for one declaration.",
"properties": {
"policy_version": {
"minLength": 1,
"title": "Policy Version",
"type": "string"
},
"anchor_block_id": {
"minLength": 1,
"title": "Anchor Block Id",
"type": "string"
},
"section_id": {
"minLength": 1,
"title": "Section Id",
"type": "string"
},
"source_order_start": {
"minimum": 0,
"title": "Source Order Start",
"type": "integer"
},
"source_order_end": {
"minimum": 0,
"title": "Source Order End",
"type": "integer"
},
"allowed_block_ids": {
"items": {
"type": "string"
},
"minItems": 1,
"title": "Allowed Block Ids",
"type": "array"
},
"explicit_backward_scope": {
"default": false,
"title": "Explicit Backward Scope",
"type": "boolean"
}
},
"required": [
"policy_version",
"anchor_block_id",
"section_id",
"source_order_start",
"source_order_end",
"allowed_block_ids"
],
"title": "DeclarationLocality",
"type": "object"
},
"DeterministicProductSignal": {
"additionalProperties": false,
"description": "Exact source span and context for a deterministic product signal.",
"properties": {
"signal_id": {
"minLength": 1,
"title": "Signal Id",
"type": "string"
},
"signal_type": {
"$ref": "#/$defs/ProductSignalType"
},
"block_id": {
"minLength": 1,
"title": "Block Id",
"type": "string"
},
"page_number": {
"minimum": 1,
"title": "Page Number",
"type": "integer"
},
"section_id": {
"minLength": 1,
"title": "Section Id",
"type": "string"
},
"literal_value": {
"title": "Literal Value",
"type": "string"
},
"normalized_value": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"title": "Normalized Value"
},
"character_start": {
"minimum": 0,
"title": "Character Start",
"type": "integer"
},
"character_end": {
"minimum": 0,
"title": "Character End",
"type": "integer"
},
"method_version": {
"minLength": 1,
"title": "Method Version",
"type": "string"
}
},
"required": [
"signal_id",
"signal_type",
"block_id",
"page_number",
"section_id",
"literal_value",
"character_start",
"character_end",
"method_version"
],
"title": "DeterministicProductSignal",
"type": "object"
},
"EvidenceEligibility": {
"description": "Product applicability eligibility before evidence selection.",
"enum": [
"eligible",
"excluded",
"review_only",
"document_context"
],
"title": "EvidenceEligibility",
"type": "string"
},
"ProductAliasProvenance": {
"additionalProperties": false,
"description": "Allowed product identity with explicit non-model provenance.",
"properties": {
"alias": {
"minLength": 1,
"title": "Alias",
"type": "string"
},
"source": {
"$ref": "#/$defs/AliasSource"
},
"derivation_method": {
"minLength": 1,
"title": "Derivation Method",
"type": "string"
},
"method_version": {
"minLength": 1,
"title": "Method Version",
"type": "string"
},
"source_block_id": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"title": "Source Block Id"
},
"character_start": {
"anyOf": [
{
"minimum": 0,
"type": "integer"
},
{
"type": "null"
}
],
"default": null,
"title": "Character Start"
},
"character_end": {
"anyOf": [
{
"minimum": 0,
"type": "integer"
},
{
"type": "null"
}
],
"default": null,
"title": "Character End"
},
"source_literal": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"title": "Source Literal"
},
"normalization_rule": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"title": "Normalization Rule"
}
},
"required": [
"alias",
"source",
"derivation_method",
"method_version"
],
"title": "ProductAliasProvenance",
"type": "object"
},
"ProductBlockDecision": {
"additionalProperties": false,
"description": "Validated authoritative linkage for one block-level scope decision.",
"properties": {
"block_id": {
"minLength": 1,
"title": "Block Id",
"type": "string"
},
"document_id": {
"minLength": 1,
"title": "Document Id",
"type": "string"
},
"page_number": {
"minimum": 1,
"title": "Page Number",
"type": "integer"
},
"section_id": {
"minLength": 1,
"title": "Section Id",
"type": "string"
},
"source_order": {
"minimum": 0,
"title": "Source Order",
"type": "integer"
},
"scope_type": {
"$ref": "#/$defs/ProductScopeType"
},
"eligibility": {
"$ref": "#/$defs/EvidenceEligibility"
},
"applicable_products": {
"items": {
"type": "string"
},
"title": "Applicable Products",
"type": "array"
},
"family_label": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"title": "Family Label"
},
"qualifying_block_ids": {
"items": {
"type": "string"
},
"title": "Qualifying Block Ids",
"type": "array"
},
"declaration_ids": {
"items": {
"type": "string"
},
"title": "Declaration Ids",
"type": "array"
},
"determination_method": {
"$ref": "#/$defs/ResolutionDeterminationMethod"
},
"rule_version": {
"minLength": 1,
"title": "Rule Version",
"type": "string"
},
"prompt_version": {
"minLength": 1,
"title": "Prompt Version",
"type": "string"
},
"rationale": {
"minLength": 1,
"title": "Rationale",
"type": "string"
},
"product_match_confidence": {
"$ref": "#/$defs/ConfidenceSignal"
},
"review_required": {
"title": "Review Required",
"type": "boolean"
},
"review_reasons": {
"items": {
"$ref": "#/$defs/ProductReviewReason"
},
"title": "Review Reasons",
"type": "array"
},
"warnings": {
"items": {
"$ref": "#/$defs/DatasetWarning"
},
"title": "Warnings",
"type": "array"
}
},
"required": [
"block_id",
"document_id",
"page_number",
"section_id",
"source_order",
"scope_type",
"eligibility",
"determination_method",
"rule_version",
"prompt_version",
"rationale",
"product_match_confidence",
"review_required"
],
"title": "ProductBlockDecision",
"type": "object"
},
"ProductDeclaration": {
"additionalProperties": false,
"description": "Auditable product or family declaration assembled from source signals.",
"properties": {
"declaration_id": {
"minLength": 1,
"title": "Declaration Id",
"type": "string"
},
"kind": {
"$ref": "#/$defs/DeclarationKind"
},
"block_id": {
"minLength": 1,
"title": "Block Id",
"type": "string"
},
"signal_ids": {
"items": {
"type": "string"
},
"minItems": 1,
"title": "Signal Ids",
"type": "array"
},
"product_values": {
"items": {
"type": "string"
},
"title": "Product Values",
"type": "array"
},
"family_label": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"title": "Family Label"
},
"target_included": {
"title": "Target Included",
"type": "boolean"
},
"contradictory": {
"default": false,
"title": "Contradictory",
"type": "boolean"
},
"method_version": {
"minLength": 1,
"title": "Method Version",
"type": "string"
},
"locality": {
"anyOf": [
{
"$ref": "#/$defs/DeclarationLocality"
},
{
"type": "null"
}
],
"default": null
}
},
"required": [
"declaration_id",
"kind",
"block_id",
"signal_ids",
"target_included",
"method_version"
],
"title": "ProductDeclaration",
"type": "object"
},
"ProductReviewReason": {
"description": "Versioned reasons for block- or dataset-level product review.",
"enum": [
"model_review_advice",
"target_block_confidence_below_threshold",
"ambiguous_product_scope",
"mixed_product_transition",
"document_target_confidence_below_threshold",
"vague_or_undetermined_coverage",
"target_inclusion_unproven",
"family_membership_unproven",
"document_contradiction",
"no_defensible_target_passage"
],
"title": "ProductReviewReason",
"type": "string"
},
"ProductScopeType": {
"description": "How an evidence block applies to products.",
"enum": [
"exact_target",
"enumerated_target",
"product_family",
"ambiguous",
"other_product",
"document_context",
"mixed_product_transition"
],
"title": "ProductScopeType",
"type": "string"
},
"ProductSignalType": {
"description": "Versioned deterministic product-applicability signal kinds.",
"enum": [
"exact_short_name",
"exact_native_id",
"exact_version",
"explicit_alias_declaration",
"enumerated_product_list",
"target_in_enumeration",
"other_product_in_enumeration",
"family_declaration",
"target_family_membership",
"unproven_family_membership",
"template_heading",
"document_wide_context",
"shared_context",
"other_product_restriction",
"unresolved_scope",
"normalized_target_alias",
"other_product_heading"
],
"title": "ProductSignalType",
"type": "string"
},
"ResolutionDeterminationMethod": {
"description": "Whether a result came from rules, model analysis, or both.",
"enum": [
"deterministic",
"model",
"hybrid"
],
"title": "ResolutionDeterminationMethod",
"type": "string"
}
},
"additionalProperties": false,
"description": "Complete blind coverage and product-scope output before evidence selection.",
"properties": {
"target": {
"$ref": "#/$defs/DatasetIdentity"
},
"extraction_manifest_sha256": {
"pattern": "^[0-9a-f]{64}$",
"title": "Extraction Manifest Sha256",
"type": "string"
},
"selected_candidate_id": {
"minLength": 1,
"title": "Selected Candidate Id",
"type": "string"
},
"selected_readme_url": {
"minLength": 1,
"title": "Selected Readme Url",
"type": "string"
},
"pdf_sha256": {
"pattern": "^[0-9a-f]{64}$",
"title": "Pdf Sha256",
"type": "string"
},
"deterministic_analysis_version": {
"minLength": 1,
"title": "Deterministic Analysis Version",
"type": "string"
},
"coverage_prompt_version": {
"enum": [
"dataset_coverage_v1",
"dataset_coverage_v2",
"dataset_coverage_v3"
],
"title": "Coverage Prompt Version",
"type": "string"
},
"product_resolution_prompt_version": {
"enum": [
"dataset_product_resolution_v1",
"dataset_product_resolution_v2",
"dataset_product_resolution_v3"
],
"title": "Product Resolution Prompt Version",
"type": "string"
},
"response_schema_version": {
"enum": [
"dataset_product_resolution_schema_v1",
"dataset_product_resolution_schema_v2",
"dataset_product_resolution_schema_v3"
],
"title": "Response Schema Version",
"type": "string"
},
"declaration_locality_policy_version": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"title": "Declaration Locality Policy Version"
},
"downstream_eligibility_policy_version": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"title": "Downstream Eligibility Policy Version"
},
"review_aggregation_policy_version": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"title": "Review Aggregation Policy Version"
},
"mixed_product_transition_policy_version": {
"anyOf": [
{
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"title": "Mixed Product Transition Policy Version"
},
"signals": {
"items": {
"$ref": "#/$defs/DeterministicProductSignal"
},
"title": "Signals",
"type": "array"
},
"declarations": {
"items": {
"$ref": "#/$defs/ProductDeclaration"
},
"title": "Declarations",
"type": "array"
},
"aliases": {
"items": {
"$ref": "#/$defs/ProductAliasProvenance"
},
"title": "Aliases",
"type": "array"
},
"coverage_mode": {
"$ref": "#/$defs/CoverageMode"
},
"qualifying_block_ids": {
"items": {
"type": "string"
},
"title": "Qualifying Block Ids",
"type": "array"
},
"rejected_block_ids": {
"items": {
"type": "string"
},
"title": "Rejected Block Ids",
"type": "array"
},
"determination_method": {
"$ref": "#/$defs/ResolutionDeterminationMethod"
},
"rationale": {
"minLength": 1,
"title": "Rationale",
"type": "string"
},
"product_match_confidence": {
"$ref": "#/$defs/ConfidenceSignal"
},
"review_required": {
"title": "Review Required",
"type": "boolean"
},
"review_reasons": {
"items": {
"$ref": "#/$defs/ProductReviewReason"
},
"title": "Review Reasons",
"type": "array"
},
"uncertain_block_ids": {
"items": {
"type": "string"
},
"title": "Uncertain Block Ids",
"type": "array"
},
"downstream_classification_eligible": {
"title": "Downstream Classification Eligible",
"type": "boolean"
},
"block_decisions": {
"items": {
"$ref": "#/$defs/ProductBlockDecision"
},
"title": "Block Decisions",
"type": "array"
},
"target_eligible_block_ids": {
"items": {
"type": "string"
},
"title": "Target Eligible Block Ids",
"type": "array"
},
"excluded_other_product_block_ids": {
"items": {
"type": "string"
},
"title": "Excluded Other Product Block Ids",
"type": "array"
},
"ambiguous_review_block_ids": {
"items": {
"type": "string"
},
"title": "Ambiguous Review Block Ids",
"type": "array"
},
"document_context_block_ids": {
"items": {
"type": "string"
},
"title": "Document Context Block Ids",
"type": "array"
},
"warnings": {
"items": {
"$ref": "#/$defs/DatasetWarning"
},
"title": "Warnings",
"type": "array"
},
"blind_cache_identity_sha256": {
"pattern": "^[0-9a-f]{64}$",
"title": "Blind Cache Identity Sha256",
"type": "string"
},
"model_calls": {
"minimum": 0,
"title": "Model Calls",
"type": "integer"
}
},
"required": [
"target",
"extraction_manifest_sha256",
"selected_candidate_id",
"selected_readme_url",
"pdf_sha256",
"deterministic_analysis_version",
"coverage_prompt_version",
"product_resolution_prompt_version",
"response_schema_version",
"signals",
"declarations",
"aliases",
"coverage_mode",
"qualifying_block_ids",
"rejected_block_ids",
"determination_method",
"rationale",
"product_match_confidence",
"review_required",
"downstream_classification_eligible",
"block_decisions",
"target_eligible_block_ids",
"excluded_other_product_block_ids",
"ambiguous_review_block_ids",
"document_context_block_ids",
"blind_cache_identity_sha256",
"model_calls"
],
"title": "DatasetCoverageResolution",
"type": "object"
}