README.md

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.

• original analogy questions

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)

• extra analogy questions

name	Size (valid/test)	Num of choice (valid/test)	Num of relation group (valid/test)	Original Reference
semeval2012_relational_similarity	79/-	3/-	79/-	relbert/semeval2012_relational_similarity
t_rex_relational_similarity	496/183	74/48	60/19	relbert/t_rex_relational_similarity
conceptnet_relational_similarity	1112/1192	19/17	18/16	relbert/conceptnet_relational_similarity
nell_relational_similarity	400/600	5/7	4/6	relbert/nell_relational_similarity
scan	178/1616	3,36,136,10,45,78,15,21,55,120,153,91,28/3,36,136,10,45,78,15,21,55,120,153,91,28	2/2	relbert/scientific_and_creative_analogy

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.