metadata

language:
  - en
license:
  - other
multilinguality:
  - monolingual
size_categories:
  - n<1K
pretty_name: Analogy Question

Dataset Card for "relbert/analogy_questions"

Dataset Description

Repository: RelBERT
Paper: https://aclanthology.org/2021.acl-long.280/
Dataset: Analogy Questions

Dataset Summary

This dataset contains 5 different word analogy questions used in Analogy Language Model.

name	Size (valid/test)	Num of choice	Num of relation group	Original Reference
`sat_full`	-/374	5	2	Turney (2005)
`sat`	37/337	5	2	Turney (2005)
`u2`	24/228	5,4,3	9	EnglishForEveryone
`u4`	48/432	5,4,3	5	EnglishForEveryone
`google`	50/500	4	2	Mikolov et al., (2013)
`bats`	199/1799	4	3	Gladkova et al., (2016)

Dataset Structure

Data Instances

An example of test looks as follows.

{
    "stem": ["raphael", "painter"],
    "answer": 2,
    "choice": [["andersen", "plato"],
                ["reading", "berkshire"],
                ["marx", "philosopher"],
                ["tolstoi", "edison"]]
}

The stem is the query word pair, choice has word pair candidates, and answer indicates the index of correct candidate which starts from 0. All data is lowercased except Google dataset.

Citation Information

@inproceedings{ushio-etal-2021-bert-is,
    title ={{BERT} is to {NLP} what {A}lex{N}et is to {CV}: {C}an {P}re-{T}rained {L}anguage {M}odels {I}dentify {A}nalogies?},
    author={Ushio, Asahi and
            Espinosa-Anke, Luis and
            Schockaert, Steven and
            Camacho-Collados, Jose},
    booktitle={Proceedings of the {ACL}-{IJCNLP} 2021 Main Conference},
    year={2021},
    publisher={Association for Computational Linguistics}
}

LICENSE

The LICENSE of all the resources are under CC-BY-NC-4.0. Thus, they are freely available for academic purpose or individual research, but restricted for commercial use.