src.dackar.RCA.cross_pattern.summary

Attributes

_OUTCOME_WORDING

Functions

format_rca_card_cross_pattern_summary(...)

Return a brief narrative string for injection into rca_card.

build_manifest_cross_pattern_summary(...)

Return structured summary for run_manifest.artifacts.

get_cross_pattern_attention_flags(candidate_evidences, ...)

Return attention flag strings for injection into analyst_attention_flags.

Module Contents

src.dackar.RCA.cross_pattern.summary._OUTCOME_WORDING: Dict[str, str][source]
src.dackar.RCA.cross_pattern.summary.format_rca_card_cross_pattern_summary(candidate_evidences, linkage_outcome_distribution)[source]

Return a brief narrative string for injection into rca_card.

Uses exact wording from §4.7 linkage_outcome table for each outcome type. When linked candidates exist, summarizes support posture across them.

Parameters:
Return type:

str

src.dackar.RCA.cross_pattern.summary.build_manifest_cross_pattern_summary(candidate_evidences, total_episodes, total_docs, total_links, links_above_threshold)[source]

Return structured summary for run_manifest.artifacts.

Includes per-candidate linkage outcome and support posture breakdowns.

Parameters:
Return type:

Dict[str, Any]

src.dackar.RCA.cross_pattern.summary.get_cross_pattern_attention_flags(candidate_evidences, candidates, top_n_candidates=3)[source]

Return attention flag strings for injection into analyst_attention_flags.

Flags raised: - “conflicting” support posture on any of top_n_candidates - “multiple_consistent” reinforcement (positive signal) on top candidates - “no_data” outcome (data coverage gap) for any candidate - “stale” index (detected via provenance on any link)

Parameters:
Return type:

List[str]