Skip to main content

agrag.ingestion.resolve

Entity resolution public API.

Modules:

  • batch_validation – Validation for LLM batch entity-match verdicts.
  • candidate_source – Candidate generation for in-batch and persisted graph entities.
  • comparators – Comparison strategies used by entity resolution.
  • exact_groups – Exact-name grouping for permanent raw entity records.
  • resolution – Resolve mentions and persisted entities against the graph.
  • resolver – Entity resolution: deciding which ExtractedEntity mentions are the same thing.
  • zone_classifier – Zone classification for entity-resolution candidate pairs.

Classes:

  • BatchResolution – The outcome of resolving one batch of mentions.
  • CandidateSource – Narrows which in-batch entity pairs resolution compares.
  • Comparator – One matching strategy a Resolver runs against a candidate pair.
  • ComparisonResult – The verdict and evidence produced by one comparator.
  • ComparisonVerdict – A Comparator's verdict on one entity pair.
  • ExactMatch – Matches when normalized text is identical. Never returns NO_MATCH.
  • FuzzyMatch – Fast-path accepter for near-identical names. Never returns NO_MATCH.
  • GraphCandidateSource – Block by label in-batch. Search persisted entities globally with ANN.
  • LLMVerify – Ask an LLM to verify an ambiguous pair. Last resort. Never UNCERTAIN.
  • PersistedCandidateSource – Supplies only candidate pairs between new mentions and raw graph entities.
  • ResolutionGroup – One set of ExtractedEntity indices resolution decided are the same entity.
  • ResolutionResult – The groups, non-exact evidence, and ambiguity count of one pass.
  • ResolvedMatch – One confirmed non-exact match between two input entity indices.
  • Resolver – Routes blocked candidate pairs through exact, fuzzy, embedding, and LLM zones.

Functions:

Attributes: