Skip to content

Dataset · gsarti

Magpie idiom corpus

gsarti/magpie

Around 37k annotated examples of potential idioms in figurative or literal contexts.

What's inside

The dataset is based on the British National Corpus (BNC) and contains between 10K and 100K English texts. It includes about 37k samples for 1,482 idioms annotated as fully figurative or literal expressions.

What you can do with it

  • Text classification and semantic analysis of idiomatic expressions
  • Researching idiom comprehension by neural machine translation models
Tasks
Text classification, Translation
Languages
English
Size
10K–100K rows
Downloads this month
704
Likes
8

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("gsarti/magpie")