Triple
T7838509
| Position | Surface form | Disambiguated ID | Type / Status |
|---|---|---|---|
| Subject | President of Afghanistan |
E181744
|
entity |
| Predicate | lastHolderEndEvent |
P33300
|
FINISHED |
| Object | Taliban offensive and fall of Kabul (2021) |
—
|
LITERAL FINISHED |
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: Taliban offensive and fall of Kabul (2021) | Statement: [President of Afghanistan, lastHolderEndEvent, Taliban offensive and fall of Kabul (2021)]
PD
Predicate disambiguation
gpt-5-mini-2025-08-07
Target predicate: lastHolderEndEvent Context triple: [President of Afghanistan, lastHolderEndEvent, Taliban offensive and fall of Kabul (2021)]
-
A.
lastHolder
Indicates that the subject is the most recent entity to have possessed, owned, or held the object.
-
B.
firstHolderEnd
Indicates that the specified entity is the one who initially held something and that this holding relationship has come to an end.
-
C.
termEndOfLastHolder
chosen
Indicates that something marks the end point of the term or tenure of the most recent previous holder of a role, position, or asset.
-
D.
endEvent
Indicates that one event or process marks the termination or conclusion of another event or process.
-
E.
lastRecordedEvent
Indicates the most recent event that has been recorded for a given entity or context.
- F. None of above.
Provenance (3 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_69ca8284a25c8190a1a20afad30da792 |
completed | March 30, 2026, 2:02 p.m. |
| NER | Named-entity recognition | batch_69cb14c336f08190a21a3fcaa7269457 |
completed | March 31, 2026, 12:26 a.m. |
| PD | Predicate disambiguation | batch_69cae91e98988190abd4ece75932c589 |
completed | March 30, 2026, 9:20 p.m. |
Created at: March 30, 2026, 4:46 p.m.