Skip to main content
Version: 0.0.5 (stable)

agrag.common.data_models.candidate.CandidateRecord

Bases: BaseModel

One rejected extraction item, stored outside the entity graph.

Each job uses a new node id. The key groups sightings across jobs without changing a committed node's count. Entity and property candidates need text. Relation endpoints need an entity id or unresolved text.

Attributes:

  • id (UUID) – Per-job node id, random unless supplied by the caller.
  • kind (CandidateKind) – Entity, relation or property candidate.
  • label (str) – Proposed entity type, relation type or property name.
  • text (str) – Entity surface text or property value. Empty for relations.
  • source_text (str | None) – Unresolved source endpoint or property holder text.
  • target_text (str | None) – Unresolved target endpoint text.
  • confidence (float) – Extractor score from 0 to 1. Use 0 for an unscored item.
  • reason (CandidateReason) – Why the extractor rejected the item.
  • chunk_id (UUID) – Evidence chunk id.
  • source_entity_id (UUID | None) – Source endpoint or property holder, when resolved.
  • target_entity_id (UUID | None) – Target endpoint, when resolved.
  • source_label (str | None) – Source endpoint or property holder type, when known.
  • target_label (str | None) – Target endpoint type, when known.
  • state (CandidateState) – Review state, initially pending.
  • count (int) – Sightings represented by this node, fixed on creation.
  • schema_fingerprint (str) – Schema used to extract the item, empty if unknown.

chunk_id​

chunk_id: UUID

confidence​

confidence: float = Field(ge=0.0, le=1.0)

count​

count: int = Field(default=1, ge=1)

evidence_edge_id​

evidence_edge_id: UUID

Return the deterministic evidence edge id.

id​

id: UUID = Field(default_factory=uuid4)

key​

key: str

Return the stable deduplication key for this record.

kind​

kind: CandidateKind

label​

label: str = Field(min_length=1)

reason​

reason: CandidateReason

schema_fingerprint​

schema_fingerprint: str = ''

source_edge_id​

source_edge_id: UUID | None

Return the deterministic source edge id when its endpoint exists.

source_entity_id​

source_entity_id: UUID | None = None

source_label​

source_label: str | None = None

source_text​

source_text: str | None = None

state​

state: CandidateState = 'pending'

target_edge_id​

target_edge_id: UUID | None

Return the deterministic target edge id when its endpoint exists.

target_entity_id​

target_entity_id: UUID | None = None

target_label​

target_label: str | None = None

target_text​

target_text: str | None = None

text​

text: str = ''