Triple
T18204574
| Position | Surface form | Disambiguated ID | Type / Status |
|---|---|---|---|
| Subject | DeBERTa |
E435870
|
entity |
| Predicate | usesPretrainingData |
P21227
|
FINISHED |
| Object | BookCorpus (for some variants) |
—
|
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: BookCorpus (for some variants) | Statement: [DeBERTa, usesPretrainingData, BookCorpus (for some variants)]
NED1
Entity disambiguation (via context triple)
gpt-5-mini-2025-08-07
Target entity: BookCorpus (for some variants) Context triple: [DeBERTa, usesPretrainingData, BookCorpus (for some variants)]
-
A.
BookCorpus
chosen
BookCorpus is a large collection of freely available books commonly used as a pretraining dataset for natural language processing models.
-
B.
Corpus
Corpus is a common shortened name for Corpus Christi College, one of the historic constituent colleges of the University of Cambridge.
-
C.
Collins Corpus
Collins Corpus is a large, computer-readable collection of real-world English texts used by Collins for corpus-based lexicography and language research.
-
D.
CORDE corpus
The CORDE corpus is a large historical Spanish language corpus compiled by the Royal Spanish Academy, used for studying the evolution and usage of Spanish over time.
-
E.
WebText dataset
The WebText dataset is a large-scale corpus of web pages curated by OpenAI to train language models like GPT-2 on diverse, high-quality internet text.
- 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_69d8b90dba6481908e119eb9aa4ca0cb |
completed | April 10, 2026, 8:47 a.m. |
| NER | Named-entity recognition | batch_69e4e222831081908f7d5500424e3acb |
completed | April 19, 2026, 2:09 p.m. |
Created at: April 10, 2026, 10:32 a.m.