Triple
T19168140
| Position | Surface form | Disambiguated ID | Type / Status |
|---|---|---|---|
| Subject | Jason Patric |
E469239
|
entity |
| Predicate | father |
P120
|
FINISHED |
| Object | Jason Miller |
—
|
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: Jason Miller | Statement: [Jason Patric, father, Jason Miller]
NED1
Entity disambiguation (via context triple)
gpt-5-mini-2025-08-07
Target entity: Jason Miller Context triple: [Jason Patric, father, Jason Miller]
-
A.
Jason Miller
chosen
Jason Miller was an American actor and playwright best known for his Oscar-nominated role as Father Damien Karras in the classic horror film "The Exorcist."
-
B.
Jason Miller
Jason Miller is a software engineer best known as the creator of the lightweight JavaScript library Preact.
-
C.
Jeff Kline
Jeff Kline is an American television producer and writer best known for his work on animated series such as Transformers: Prime.
-
D.
Alan Ruck
Alan Ruck is an American actor best known for his roles as Cameron Frye in "Ferris Bueller's Day Off" and Connor Roy in the television series "Succession."
-
E.
Timothy Busfield
Timothy Busfield is an American actor and director best known for his roles in television series such as "thirtysomething," "The West Wing," and various film and stage productions.
- 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_69d8dd09d5a081909ae43c286651ae5a |
completed | April 10, 2026, 11:20 a.m. |
| NER | Named-entity recognition | batch_69e5f16132e081908c6b8d576163316e |
completed | April 20, 2026, 9:26 a.m. |
Created at: April 10, 2026, 12:06 p.m.