Dataset Viewer

The dataset viewer is not available because its heuristics could not detect any supported data files. You can try uploading some data files, or configuring the data files location manually.

thunlp/few_rel(複製)

データの内容

100 を超える関係を持つ、英語の少数ショット関係抽出のデータ。1 件は 1 文(tokens)と、その中の 2 つの実体(head と tail、語の位置と型を含む)、関係の ID(relation)からなる。文は Wikipedia の執筆者が書いた英語で、アノテーションはクラウドソーシングで作られた。一般的な話題のデータで、金融に特化していない。

データの概要

項目 値
行数 71,595(2 構成・7 分割の合計)
列数 2〜5(構成による)
容量 17.15 KB(ファイル 3 本)
ファイルの形式 py 1
言語 en
公開 public

構成と分割

構成(config) 分割(split)ごとの行数 列数
default train_wiki 44,800、val_nyt 2,500、val_pubmed 1,000、val_semeval 8,851、val_wiki 11,200、pubmed_unsupervised 2,500 5
pid2name pid2name 744 2

列(構成 default。構成ごとに列が違う)

列 型
relation string
tokens list<string>
head struct{text, type, indices}
tail struct{text, type, indices}
names list<string>

行数と列は元のカードの YAML(dataset_info)の記載。データセットビューアでは読めなかった。全データセットの一覧は HYUKI5/dataset-catalog にある。

複製について

thunlp/few_rel をリビジョン 12d1071d9cb5b7526c81a2784be3716acf8c6c00 の時点でそのまま複製したもの。中身は変えていない。判断モデル Laya(convaiinnovations/laya)の微調整に使う候補として、調査メモ(Laya 学習用データセット調査)に載せたデータを集めている。

出典

論文

アノテーションの方法

データセットの作成者

FewRel の場合:

Han, Xu and Zhu, Hao and Yu, Pengfei and Wang, Ziyun and Yao, Yuan and Liu, Zhiyuan and Sun, Maosong

FewRel 2.0 の場合:

Gao, Tianyu and Han, Xu and Zhu, Hao and Liu, Zhiyuan and Li, Peng and Sun, Maosong and Zhou, Jie

微調整に使うときの注意

  • 調査メモの台帳での区分: D(記載なし・other・unknown・資料間の食い違い)。区分は記載の文言で機械的に付けたもので、法的な判定ではない。学習に入れる前に元のカードとライセンスの原文で確かめる。
  • Laya の入力は 2,048 token まで(ModernBERT-large / mmBERT)。長い本文は先頭から黙って切られるので、切れる行は事前に除くか、要約・切り出しをしてから入れる。
  • 分割は記事・文書単位で行い、同じ本文から作った別の問いを学習側と較正側に分けない。温度と閾値は学習に使っていない分割で別々に当てる。

調査メモの記述

メモ 節 ライセンス(記載) 確認 注記
ontology_kg_entity.md Q1. 関係分類(TACRED 系・DocRED・FewRel・SemEval-2010・金融関係抽出)にどんなデータがあり、「供給・競合・買収・提携」を判断する Laya にどう載るか (L30) thunlp: MIT。tasksource: unknown 確認済 複数/混在 / 確認印なし(各ファイルの凡例により確認済扱い) / 原文注記: 一般ドメインで金融特化でない。スクリプト版は datasets 4 以降で読めない可能性(general_english_nlu.md §2 の同種の指摘)。tasksource の parquet を優先

元のカード(日本語訳)

以下は元のデータセットのカードの本文を日本語に訳したもの(機械翻訳)。訳に疑問があるときは、末尾の原文を確かめる。


few_rel のデータセットカード

目次

  • データセットの説明
    • データセットの概要
    • 対応タスクとリーダーボード
    • 言語
  • データセットの構造
    • データインスタンス
    • データフィールド
    • データ分割
  • データセットの作成
    • キュレーションの根拠
    • 元データ
    • アノテーション
    • 個人情報と機微な情報
  • データの利用上の留意点
    • データセットの社会的影響
    • バイアスに関する議論
    • その他の既知の制限
  • 追加情報
    • データセットの作成者
    • ライセンス情報
    • 引用情報
    • 貢献者

データセットの説明

データセットの概要

FewRel は、100 を超える関係と、異なるドメインにわたる数万件のアノテーション付きインスタンスを含む、大規模な少数ショット関係抽出データセットである。

対応タスクとリーダーボード

[追加情報が必要]

言語

本データセットには、Wikipedia の執筆者が使う英語のテキストと、クラウドソーシングで作られた英語のアノテーションが含まれる。

データセットの構造

データインスタンス

train_wiki split からのインスタンス:

{'head': {'indices': [[16]], 'text': 'tjq', 'type': 'Q1331049'}, 'names': ['place served by transport hub', 'territorial entity or entities served by this transport hub (airport, train station, etc.)'], 'relation': 'P931', 'tail': {'indices': [[13, 14]], 'text': 'tanjung pandan', 'type': 'Q3056359'}, 'tokens': ['Merpati', 'flight', '106', 'departed', 'Jakarta', '(', 'CGK', ')', 'on', 'a', 'domestic', 'flight', 'to', 'Tanjung', 'Pandan', '(', 'TJQ', ')', '.']}

データフィールド

default の場合:

  • relation: 関係の PID を含む string の特徴量。

  • tokens: テキストのトークンを含む string の特徴量の list。

  • head: 次を含む辞書:

    • text: head エンティティを表す string の特徴量。
    • type: head エンティティの型を表す string の特徴量。
    • indices: トークンのインデックスの list を含む list。
  • tail: 次を含む辞書:

    • text: tail エンティティを表す string の特徴量。
    • type: tail エンティティの型を表す string の特徴量。
    • indices: トークンのインデックスの list を含む list。
  • names: 関係名を含む string の特徴量の list。pubmed_unsupervised split では、空の string を持つ list に設定される。val_semeval と val_pubmed split では、relation フィールドの string を持つ list に設定される。

データ分割

train_wiki: 44800 val_nyt: 2500 val_pubmed: 1000 val_semeval: 8851 val_wiki: 11200 pubmed_unsupervised: 2500

データセットの作成

キュレーションの根拠

[追加情報が必要]

元データ

初期データの収集と正規化

[追加情報が必要]

原文の言語の作成者は誰か?

[追加情報が必要]

アノテーション

アノテーションの過程

[追加情報が必要]

アノテーターは誰か?

[追加情報が必要]

個人情報と機微な情報

[追加情報が必要]

データの利用上の留意点

データセットの社会的影響

[追加情報が必要]

バイアスに関する議論

[追加情報が必要]

その他の既知の制限

[追加情報が必要]

追加情報

データセットの作成者

FewRel の場合:

Han, Xu and Zhu, Hao and Yu, Pengfei and Wang, Ziyun and Yao, Yuan and Liu, Zhiyuan and Sun, Maosong

FewRel 2.0 の場合:

Gao, Tianyu and Han, Xu and Zhu, Hao and Liu, Zhiyuan and Li, Peng and Sun, Maosong and Zhou, Jie

ライセンス情報

MIT License

Copyright (c) 2018 THUNLP

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.

引用情報

@inproceedings{han-etal-2018-fewrel,
    title = "{F}ew{R}el: A Large-Scale Supervised Few-Shot Relation Classification Dataset with State-of-the-Art Evaluation",
    author = "Han, Xu and Zhu, Hao and Yu, Pengfei and Wang, Ziyun and Yao, Yuan and Liu, Zhiyuan and Sun, Maosong",
    booktitle = "Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing",
    month = oct # "-" # nov,
    year = "2018",
    address = "Brussels, Belgium",
    publisher = "Association for Computational Linguistics",
    url = "https://www.aclweb.org/anthology/D18-1514",
    doi = "10.18653/v1/D18-1514",
    pages = "4803--4809"
}

@inproceedings{gao-etal-2019-fewrel,
    title = "{F}ew{R}el 2.0: Towards More Challenging Few-Shot Relation Classification",
    author = "Gao, Tianyu and Han, Xu and Zhu, Hao and Liu, Zhiyuan and Li, Peng and Sun, Maosong and Zhou, Jie",
    booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP)",
    month = nov,
    year = "2019",
    address = "Hong Kong, China",
    publisher = "Association for Computational Linguistics",
    url = "https://www.aclweb.org/anthology/D19-1649",
    doi = "10.18653/v1/D19-1649",
    pages = "6251--6256"
}

貢献者

このデータセットを追加してくれた @gchhablani に感謝する。

原文(英語)

Dataset Card for few_rel

Dataset Summary

FewRel is a large-scale few-shot relation extraction dataset, which contains more than one hundred relations and tens of thousands of annotated instances cross different domains.

Supported Tasks and Leaderboards

[Needs More Information]

Languages

The dataset contaings English text, as used by writers on Wikipedia, and crowdsourced English annotations.

Dataset Structure

Data Instances

An instance from train_wiki split:

{'head': {'indices': [[16]], 'text': 'tjq', 'type': 'Q1331049'}, 'names': ['place served by transport hub', 'territorial entity or entities served by this transport hub (airport, train station, etc.)'], 'relation': 'P931', 'tail': {'indices': [[13, 14]], 'text': 'tanjung pandan', 'type': 'Q3056359'}, 'tokens': ['Merpati', 'flight', '106', 'departed', 'Jakarta', '(', 'CGK', ')', 'on', 'a', 'domestic', 'flight', 'to', 'Tanjung', 'Pandan', '(', 'TJQ', ')', '.']}

Data Fields

For default:

  • relation: a string feature containing PID of the relation.

  • tokens: a list of string features containing tokens for the text.

  • head: a dictionary containing:

    • text: a string feature representing the head entity.
    • type: a string feature representing the type of the head entity.
    • indices: a list containing list of token indices.
  • tail: a dictionary containing:

    • text: a string feature representing the tail entity.
    • type: a string feature representing the type of the tail entity.
    • indices: a list containing list of token indices.
  • names: a list of string features containing relation names. For pubmed_unsupervised split, this is set to a list with an empty string. For val_semeval and val_pubmed split, this is set to a list with the string from the relation field.

Data Splits

train_wiki: 44800 val_nyt: 2500 val_pubmed: 1000 val_semeval: 8851 val_wiki: 11200 pubmed_unsupervised: 2500

Dataset Creation

Curation Rationale

[Needs More Information]

Source Data

Initial Data Collection and Normalization

[Needs More Information]

Who are the source language producers?

[Needs More Information]

Annotations

Annotation process

[Needs More Information]

Who are the annotators?

[Needs More Information]

Personal and Sensitive Information

[Needs More Information]

Considerations for Using the Data

Social Impact of Dataset

[Needs More Information]

Discussion of Biases

[Needs More Information]

Other Known Limitations

[Needs More Information]

Additional Information

Dataset Curators

For FewRel:

Han, Xu and Zhu, Hao and Yu, Pengfei and Wang, Ziyun and Yao, Yuan and Liu, Zhiyuan and Sun, Maosong

For FewRel 2.0:

Gao, Tianyu and Han, Xu and Zhu, Hao and Liu, Zhiyuan and Li, Peng and Sun, Maosong and Zhou, Jie

Licensing Information

MIT License

Copyright (c) 2018 THUNLP

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.

Citation Information

@inproceedings{han-etal-2018-fewrel,
    title = "{F}ew{R}el: A Large-Scale Supervised Few-Shot Relation Classification Dataset with State-of-the-Art Evaluation",
    author = "Han, Xu and Zhu, Hao and Yu, Pengfei and Wang, Ziyun and Yao, Yuan and Liu, Zhiyuan and Sun, Maosong",
    booktitle = "Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing",
    month = oct # "-" # nov,
    year = "2018",
    address = "Brussels, Belgium",
    publisher = "Association for Computational Linguistics",
    url = "https://www.aclweb.org/anthology/D18-1514",
    doi = "10.18653/v1/D18-1514",
    pages = "4803--4809"
}

@inproceedings{gao-etal-2019-fewrel,
    title = "{F}ew{R}el 2.0: Towards More Challenging Few-Shot Relation Classification",
    author = "Gao, Tianyu and Han, Xu and Zhu, Hao and Liu, Zhiyuan and Li, Peng and Sun, Maosong and Zhou, Jie",
    booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP)",
    month = nov,
    year = "2019",
    address = "Hong Kong, China",
    publisher = "Association for Computational Linguistics",
    url = "https://www.aclweb.org/anthology/D19-1649",
    doi = "10.18653/v1/D19-1649",
    pages = "6251--6256"
}

Contributions

Thanks to @gchhablani for adding this dataset.

Downloads last month
85

Papers for HYUKI5/thunlp__few_rel