src.dackar.RCA.cross_pattern.models

Classes

HistoricalDocExtraction	Canonical cross-pattern representation of a single doc extraction record.
CrossPatternLink	A single link between one HistoricalSignalEpisode and one HistoricalDocExtraction.
CandidateCrossPatternEvidence	Aggregated cross-pattern evidence for a single causality candidate.

Module Contents

class src.dackar.RCA.cross_pattern.models.HistoricalDocExtraction

Canonical cross-pattern representation of a single doc extraction record.

Produced by converting a SemanticMatch (from DocExtractionStore.query()) into a richer object that carries temporal and FM resolution metadata needed for the linkage rules in rules.py.

Instances are treated as immutable inputs by CrossPatternLinker.run(): the linker never writes back onto them. Episode↔doc linkage is reported in the run() result (per-candidate linked_episode_ids / linked_doc_ids and the flat all_links), not by mutating these objects.

doc_id: str

doc_type: str

asset_id: str | None

event_time_start: datetime.datetime | None

event_time_end: datetime.datetime | None

event_time_confidence: str

identified_effect: str | None

assessed_cause: str | None

inferred_fm_label: str | None

fm_id_candidate: str | None

fm_id_candidate_alt: str | None

fm_resolution_status: str

fm_resolution_score: float | None

confidence: str

cause_is_symptom: bool

epistemic_class: str | None = None

classification_resolution_level: str | None = None

degraded_classification: bool = False

class src.dackar.RCA.cross_pattern.models.CrossPatternLink

A single link between one HistoricalSignalEpisode and one HistoricalDocExtraction.

Carries full provenance so analysts can trace exactly which checks passed (direct reference, temporal, or semantic/FM; asset compatibility is scored separately as asset_match).

link_id: str

episode_id: str

doc_id: str

asset_match: bool

time_overlap_hours: float | None

temporal_link_skipped: bool

linkage_precedence_level: int

component_overlap: List[str]

fm_alignment_score: float | None

signal_similarity_score: float

document_similarity_score: float | None

link_confidence: float

provenance: Dict[str, Any]

class src.dackar.RCA.cross_pattern.models.CandidateCrossPatternEvidence

Aggregated cross-pattern evidence for a single causality candidate.

Produced by CrossPatternLinker.run() — one instance per candidate in causality_candidates[“candidates”].

candidate_id: str

component_id: str

fm_id: str

linked_episode_ids: List[str]

linked_doc_ids: List[str]

best_link_score: float

support_posture: str

reinforcement_strength: str | None

linkage_outcome: str

evidence_paths: List[CrossPatternLink]