Triple

T20103894
Position Surface form Disambiguated ID Type / Status
Subject Breach E496617 entity
Predicate starredActor P5563 FINISHED
Object Caroline Dhavernas NE NERFINISHED

How this triple was built (2 steps)

Every LLM step that produced this triple, in pipeline order — named-entity classification, the disambiguation choices (the exact options shown, with the pick highlighted), and the generated description. The batch + timestamp of each is in the Provenance table below.

NER Named-entity recognition gpt-5-mini
Instruction
Given a phrase, classify it is english named entity (e.g., persons, organizations, works of art) in Latin script, or not (e.g., literals, dates, URLs, verbose phrases). For disambiguation, the statement where the phrase occurs as object is also given. Please return a JSON object with `phrase` (string, the phrase being analyzed) and `is_ne` (boolean, indicating whether the phrase is a Named Entity).
Input
Phrase: Caroline Dhavernas | Statement: [Breach, starredActor, Caroline Dhavernas]
NED1 Entity disambiguation (via context triple) gpt-5-mini-2025-08-07
Target entity: Caroline Dhavernas
Context triple: [Breach, starredActor, Caroline Dhavernas]
  • A. Caroline Dhavernas chosen
    Caroline Dhavernas is a Canadian actress best known for her roles in television series such as "Wonderfalls" and "Hannibal."
  • B. Caroline Cellier
    Caroline Cellier was a French film and television actress known for her versatile performances in comedies and dramas from the 1960s onward.
  • C. Caroline Lejeune
    Caroline Lejeune is a British film critic and journalist known for her influential mid-20th-century reviews and writings on cinema.
  • D. Caroline Boujard
    Caroline Boujard is a French rugby union player known for her performances as a back three player for the France women's national team and in top-level club competition.
  • E. Caroline Hervé
    Caroline Hervé is a French DJ, producer, and singer best known under her stage name Miss Kittin, a prominent figure in the electroclash and electronic music scenes.
  • F. None of above.
  • G. Unsure - the case is ambiguous/there is not enough information to decide.

Provenance (2 batches)

The batch behind each pipeline step, in order, with when it ran. Timestamps are batch-level — stages were processed in waves, so the object chain (NER → NED1 → NEDg → NED2) reads in order, but predicate / elicitation batches can sit in a different wave.

Step Stage Batch ID Status When
creating Elicitation batch_69da626eee3881909f3454986d4a6511 completed April 11, 2026, 3:02 p.m.
NER Named-entity recognition batch_69e666daf73c819089f02ca6faa2c283 completed April 20, 2026, 5:48 p.m.
Created at: April 11, 2026, 11:27 p.m.