HYUKI5/mteb__multilingual-scala-classification

mteb/multilingual-scala-classification（複製）

データの内容

デンマーク語、ノルウェー語 2 種、スウェーデン語の文に、言語的な容認可能性のラベルを付けたデータ。1 行が 1 件の文で、本文は text、corruption_type、ラベルは label に入る。Universal Dependencies の依存関係アノテーションから自動で作られ、ScandEval の一部として公開された。

データの概要

項目	値
行数	74,846（4 構成・16 分割の合計）
列数	3
容量	5.78 MB（ファイル 19 本）
ファイルの形式	parquet 16、yaml 1
言語	dan, nno, nob, swe
公開	public

構成と分割

構成（config）	分割（split）ごとの行数	列数
Danish	train 1,024、test 2,048、full_train 5,342、val 256	3
Norwegian_b	train 1,024、test 2,048、full_train 25,946、val 256	3
Norwegian_n	train 1,024、test 2,048、full_train 22,800、val 256	3
Swedish	train 1,024、test 2,048、full_train 7,446、val 256	3

列（構成 Danish・分割 train。どの構成も同じ列）

列	型	先頭の行の値
text	string	Et kort øjeblik frygtede han, at bedstefar Jonas…
corruption_type	string	null
label	string	correct

行数と列は HF のデータセットビューアの集計（元のデータセットの現在の版で数えたもの）。全データセットの一覧は HYUKI5/dataset-catalog にある。

複製について

mteb/multilingual-scala-classification をリビジョン 7f10a5bfb3610db4444b451e4f295fae99108f69 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

• 元のデータセット: https://huggingface.co/datasets/mteb/multilingual-scala-classification
• 複製したリビジョン: 7f10a5bfb3610db4444b451e4f295fae99108f69（2026-02-24 21:18:30+00:00）
• 元のライセンス表記: cc-by-sa-4.0

論文

• https://arxiv.org/abs/2502.13595
• https://arxiv.org/abs/2210.07316
• https://aclanthology.org/2023.nodalida-1.20/
• https://aclanthology.org/2023.nodalida-1.20

アノテーションの方法

（元のカードに該当の見出しが無いので、ラベルの付け方に触れた段落を抜き出した）

ScaLa は、Universal Dependencies Treebanks の依存関係アノテーションから自動的に構築された、大陸スカンジナビア諸語のための言語的容認可能性のデータセットである。 「ScandEval: A Benchmark for Scandinavian Natural Language Processing」の一部として公開された。

---

このデータセットカードは MTEB を使って自動生成された。

微調整に使うときの注意

• 調査メモの台帳での区分: B（継承条件・gated・モデル由来の条件・上流の規約などが付く）。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
multilingual_cjk.md	5. その他の言語(広がり確保用) (L259)	cc-by-sa-4.0	確認済	確認印なし(各ファイルの凡例により確認済扱い)

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。

---

ScalaClassification

MTEB のデータセット

Massive Text Embedding Benchmark

ScaLa は、Universal Dependencies Treebanks の依存関係アノテーションから自動的に構築された、大陸スカンジナビア諸語のための言語的容認可能性のデータセットである。 「ScandEval: A Benchmark for Scandinavian Natural Language Processing」の一部として公開された。

タスクのカテゴリ	t2c
ドメイン	フィクション、ニュース、ノンフィクション、ブログ、話し言葉、Web、書き言葉
参考文献	https://aclanthology.org/2023.nodalida-1.20/

このタスクでの評価方法

次のコードを使うと、このデータセットで embedding モデルを評価できる。

import mteb

task = mteb.get_tasks(["ScalaClassification"])
evaluator = mteb.MTEB(task)

model = mteb.get_model(YOUR_MODEL)
evaluator.run(model)

mteb のタスクでモデルを実行する方法の詳細は、GitHub のリポジトリ を参照すること。

引用

このデータセットを使用する場合は、データセットに加えて mteb も引用すること。このデータセットは MMTEB Contribution の一部としての追加の処理を含んでいる可能性が高い。

@inproceedings{nielsen-2023-scandeval,
  address = {T{\'o}rshavn, Faroe Islands},
  author = {Nielsen, Dan},
  booktitle = {Proceedings of the 24th Nordic Conference on Computational Linguistics (NoDaLiDa)},
  editor = {Alum{\"a}e, Tanel  and
Fishel, Mark},
  month = may,
  pages = {185--201},
  publisher = {University of Tartu Library},
  title = {{S}cand{E}val: A Benchmark for {S}candinavian Natural Language Processing},
  url = {https://aclanthology.org/2023.nodalida-1.20},
  year = {2023},
}


@article{enevoldsen2025mmtebmassivemultilingualtext,
  title={MMTEB: Massive Multilingual Text Embedding Benchmark},
  author={Kenneth Enevoldsen and Isaac Chung and Imene Kerboua and Márton Kardos and Ashwin Mathur and David Stap and Jay Gala and Wissam Siblini and Dominik Krzemiński and Genta Indra Winata and Saba Sturua and Saiteja Utpala and Mathieu Ciancone and Marion Schaeffer and Gabriel Sequeira and Diganta Misra and Shreeya Dhakal and Jonathan Rystrøm and Roman Solomatin and Ömer Çağatan and Akash Kundu and Martin Bernstorff and Shitao Xiao and Akshita Sukhlecha and Bhavish Pahwa and Rafał Poświata and Kranthi Kiran GV and Shawon Ashraf and Daniel Auras and Björn Plüster and Jan Philipp Harries and Loïc Magne and Isabelle Mohr and Mariya Hendriksen and Dawei Zhu and Hippolyte Gisserot-Boukhlef and Tom Aarsen and Jan Kostkan and Konrad Wojtasik and Taemin Lee and Marek Šuppa and Crystina Zhang and Roberta Rocca and Mohammed Hamdy and Andrianos Michail and John Yang and Manuel Faysse and Aleksei Vatolin and Nandan Thakur and Manan Dey and Dipam Vasani and Pranjal Chitale and Simone Tedeschi and Nguyen Tai and Artem Snegirev and Michael Günther and Mengzhou Xia and Weijia Shi and Xing Han Lù and Jordan Clive and Gayatri Krishnakumar and Anna Maksimova and Silvan Wehrli and Maria Tikhonova and Henil Panchal and Aleksandr Abramov and Malte Ostendorff and Zheng Liu and Simon Clematide and Lester James Miranda and Alena Fenogenova and Guangyu Song and Ruqiya Bin Safi and Wen-Ding Li and Alessia Borghini and Federico Cassano and Hongjin Su and Jimmy Lin and Howard Yen and Lasse Hansen and Sara Hooker and Chenghao Xiao and Vaibhav Adlakha and Orion Weller and Siva Reddy and Niklas Muennighoff},
  publisher = {arXiv},
  journal={arXiv preprint arXiv:2502.13595},
  year={2025},
  url={https://arxiv.org/abs/2502.13595},
  doi = {10.48550/arXiv.2502.13595},
}

@article{muennighoff2022mteb,
  author = {Muennighoff, Niklas and Tazi, Nouamane and Magne, Lo{\"\i}c and Reimers, Nils},
  title = {MTEB: Massive Text Embedding Benchmark},
  publisher = {arXiv},
  journal={arXiv preprint arXiv:2210.07316},
  year = {2022}
  url = {https://arxiv.org/abs/2210.07316},
  doi = {10.48550/ARXIV.2210.07316},
}

データセットの統計

データセットの統計

以下のコードには、このタスクの記述統計が含まれる。次のコードでも取得できる。

import mteb

task = mteb.get_task("ScalaClassification")

desc_stats = task.metadata.descriptive_stats

{
    "test": {
        "num_samples": 8192,
        "number_of_characters": 839257,
        "number_texts_intersect_with_train": 0,
        "min_text_length": 13,
        "average_text_length": 102.4483642578125,
        "max_text_length": 613,
        "unique_text": 8192,
        "unique_labels": 2,
        "labels": {
            "0": {
                "count": 4096
            },
            "1": {
                "count": 4096
            }
        },
        "hf_subset_descriptive_stats": {
            "Danish": {
                "num_samples": 2048,
                "number_of_characters": 224132,
                "number_texts_intersect_with_train": 0,
                "min_text_length": 13,
                "average_text_length": 109.439453125,
                "max_text_length": 443,
                "unique_text": 2048,
                "unique_labels": 2,
                "labels": {
                    "0": {
                        "count": 1024
                    },
                    "1": {
                        "count": 1024
                    }
                }
            },
            "Norwegian_b": {
                "num_samples": 2048,
                "number_of_characters": 201596,
                "number_texts_intersect_with_train": 0,
                "min_text_length": 18,
                "average_text_length": 98.435546875,
                "max_text_length": 397,
                "unique_text": 2048,
                "unique_labels": 2,
                "labels": {
                    "1": {
                        "count": 1024
                    },
                    "0": {
                        "count": 1024
                    }
                }
            },
            "Norwegian_n": {
                "num_samples": 2048,
                "number_of_characters": 212059,
                "number_texts_intersect_with_train": 0,
                "min_text_length": 18,
                "average_text_length": 103.54443359375,
                "max_text_length": 349,
                "unique_text": 2048,
                "unique_labels": 2,
                "labels": {
                    "1": {
                        "count": 1024
                    },
                    "0": {
                        "count": 1024
                    }
                }
            },
            "Swedish": {
                "num_samples": 2048,
                "number_of_characters": 201470,
                "number_texts_intersect_with_train": 0,
                "min_text_length": 17,
                "average_text_length": 98.3740234375,
                "max_text_length": 613,
                "unique_text": 2048,
                "unique_labels": 2,
                "labels": {
                    "1": {
                        "count": 1024
                    },
                    "0": {
                        "count": 1024
                    }
                }
            }
        }
    },
    "train": {
        "num_samples": 4096,
        "number_of_characters": 421198,
        "number_texts_intersect_with_train": null,
        "min_text_length": 14,
        "average_text_length": 102.83154296875,
        "max_text_length": 402,
        "unique_text": 4096,
        "unique_labels": 2,
        "labels": {
            "1": {
                "count": 2048
            },
            "0": {
                "count": 2048
            }
        },
        "hf_subset_descriptive_stats": {
            "Danish": {
                "num_samples": 1024,
                "number_of_characters": 110271,
                "number_texts_intersect_with_train": null,
                "min_text_length": 14,
                "average_text_length": 107.6865234375,
                "max_text_length": 392,
                "unique_text": 1024,
                "unique_labels": 2,
                "labels": {
                    "1": {
                        "count": 512
                    },
                    "0": {
                        "count": 512
                    }
                }
            },
            "Norwegian_b": {
                "num_samples": 1024,
                "number_of_characters": 97878,
                "number_texts_intersect_with_train": null,
                "min_text_length": 18,
                "average_text_length": 95.583984375,
                "max_text_length": 350,
                "unique_text": 1024,
                "unique_labels": 2,
                "labels": {
                    "1": {
                        "count": 512
                    },
                    "0": {
                        "count": 512
                    }
                }
            },
            "Norwegian_n": {
                "num_samples": 1024,
                "number_of_characters": 107913,
                "number_texts_intersect_with_train": null,
                "min_text_length": 20,
                "average_text_length": 105.3837890625,
                "max_text_length": 402,
                "unique_text": 1024,
                "unique_labels": 2,
                "labels": {
                    "1": {
                        "count": 512
                    },
                    "0": {
                        "count": 512
                    }
                }
            },
            "Swedish": {
                "num_samples": 1024,
                "number_of_characters": 105136,
                "number_texts_intersect_with_train": null,
                "min_text_length": 19,
                "average_text_length": 102.671875,
                "max_text_length": 326,
                "unique_text": 1024,
                "unique_labels": 2,
                "labels": {
                    "1": {
                        "count": 512
                    },
                    "0": {
                        "count": 512
                    }
                }
            }
        }
    }
}

---

このデータセットカードは MTEB を使って自動生成された。

原文（英語）

ScalaClassification

An MTEB dataset

Massive Text Embedding Benchmark

ScaLa a linguistic acceptability dataset for the mainland Scandinavian languages automatically constructed from dependency annotations in Universal Dependencies Treebanks. Published as part of 'ScandEval: A Benchmark for Scandinavian Natural Language Processing'

Task category	t2c
Domains	Fiction, News, Non-fiction, Blog, Spoken, Web, Written
Reference	https://aclanthology.org/2023.nodalida-1.20/

How to evaluate on this task

You can evaluate an embedding model on this dataset using the following code:

import mteb

task = mteb.get_tasks(["ScalaClassification"])
evaluator = mteb.MTEB(task)

model = mteb.get_model(YOUR_MODEL)
evaluator.run(model)

To learn more about how to run models on mteb task check out the GitHub repitory.

Citation

If you use this dataset, please cite the dataset as well as mteb, as this dataset likely includes additional processing as a part of the MMTEB Contribution.

@inproceedings{nielsen-2023-scandeval,
  address = {T{\'o}rshavn, Faroe Islands},
  author = {Nielsen, Dan},
  booktitle = {Proceedings of the 24th Nordic Conference on Computational Linguistics (NoDaLiDa)},
  editor = {Alum{\"a}e, Tanel  and
Fishel, Mark},
  month = may,
  pages = {185--201},
  publisher = {University of Tartu Library},
  title = {{S}cand{E}val: A Benchmark for {S}candinavian Natural Language Processing},
  url = {https://aclanthology.org/2023.nodalida-1.20},
  year = {2023},
}


@article{enevoldsen2025mmtebmassivemultilingualtext,
  title={MMTEB: Massive Multilingual Text Embedding Benchmark},
  author={Kenneth Enevoldsen and Isaac Chung and Imene Kerboua and Márton Kardos and Ashwin Mathur and David Stap and Jay Gala and Wissam Siblini and Dominik Krzemiński and Genta Indra Winata and Saba Sturua and Saiteja Utpala and Mathieu Ciancone and Marion Schaeffer and Gabriel Sequeira and Diganta Misra and Shreeya Dhakal and Jonathan Rystrøm and Roman Solomatin and Ömer Çağatan and Akash Kundu and Martin Bernstorff and Shitao Xiao and Akshita Sukhlecha and Bhavish Pahwa and Rafał Poświata and Kranthi Kiran GV and Shawon Ashraf and Daniel Auras and Björn Plüster and Jan Philipp Harries and Loïc Magne and Isabelle Mohr and Mariya Hendriksen and Dawei Zhu and Hippolyte Gisserot-Boukhlef and Tom Aarsen and Jan Kostkan and Konrad Wojtasik and Taemin Lee and Marek Šuppa and Crystina Zhang and Roberta Rocca and Mohammed Hamdy and Andrianos Michail and John Yang and Manuel Faysse and Aleksei Vatolin and Nandan Thakur and Manan Dey and Dipam Vasani and Pranjal Chitale and Simone Tedeschi and Nguyen Tai and Artem Snegirev and Michael Günther and Mengzhou Xia and Weijia Shi and Xing Han Lù and Jordan Clive and Gayatri Krishnakumar and Anna Maksimova and Silvan Wehrli and Maria Tikhonova and Henil Panchal and Aleksandr Abramov and Malte Ostendorff and Zheng Liu and Simon Clematide and Lester James Miranda and Alena Fenogenova and Guangyu Song and Ruqiya Bin Safi and Wen-Ding Li and Alessia Borghini and Federico Cassano and Hongjin Su and Jimmy Lin and Howard Yen and Lasse Hansen and Sara Hooker and Chenghao Xiao and Vaibhav Adlakha and Orion Weller and Siva Reddy and Niklas Muennighoff},
  publisher = {arXiv},
  journal={arXiv preprint arXiv:2502.13595},
  year={2025},
  url={https://arxiv.org/abs/2502.13595},
  doi = {10.48550/arXiv.2502.13595},
}

@article{muennighoff2022mteb,
  author = {Muennighoff, Niklas and Tazi, Nouamane and Magne, Lo{\"\i}c and Reimers, Nils},
  title = {MTEB: Massive Text Embedding Benchmark},
  publisher = {arXiv},
  journal={arXiv preprint arXiv:2210.07316},
  year = {2022}
  url = {https://arxiv.org/abs/2210.07316},
  doi = {10.48550/ARXIV.2210.07316},
}

Dataset Statistics

Dataset Statistics

The following code contains the descriptive statistics from the task. These can also be obtained using:

import mteb

task = mteb.get_task("ScalaClassification")

desc_stats = task.metadata.descriptive_stats

{
    "test": {
        "num_samples": 8192,
        "number_of_characters": 839257,
        "number_texts_intersect_with_train": 0,
        "min_text_length": 13,
        "average_text_length": 102.4483642578125,
        "max_text_length": 613,
        "unique_text": 8192,
        "unique_labels": 2,
        "labels": {
            "0": {
                "count": 4096
            },
            "1": {
                "count": 4096
            }
        },
        "hf_subset_descriptive_stats": {
            "Danish": {
                "num_samples": 2048,
                "number_of_characters": 224132,
                "number_texts_intersect_with_train": 0,
                "min_text_length": 13,
                "average_text_length": 109.439453125,
                "max_text_length": 443,
                "unique_text": 2048,
                "unique_labels": 2,
                "labels": {
                    "0": {
                        "count": 1024
                    },
                    "1": {
                        "count": 1024
                    }
                }
            },
            "Norwegian_b": {
                "num_samples": 2048,
                "number_of_characters": 201596,
                "number_texts_intersect_with_train": 0,
                "min_text_length": 18,
                "average_text_length": 98.435546875,
                "max_text_length": 397,
                "unique_text": 2048,
                "unique_labels": 2,
                "labels": {
                    "1": {
                        "count": 1024
                    },
                    "0": {
                        "count": 1024
                    }
                }
            },
            "Norwegian_n": {
                "num_samples": 2048,
                "number_of_characters": 212059,
                "number_texts_intersect_with_train": 0,
                "min_text_length": 18,
                "average_text_length": 103.54443359375,
                "max_text_length": 349,
                "unique_text": 2048,
                "unique_labels": 2,
                "labels": {
                    "1": {
                        "count": 1024
                    },
                    "0": {
                        "count": 1024
                    }
                }
            },
            "Swedish": {
                "num_samples": 2048,
                "number_of_characters": 201470,
                "number_texts_intersect_with_train": 0,
                "min_text_length": 17,
                "average_text_length": 98.3740234375,
                "max_text_length": 613,
                "unique_text": 2048,
                "unique_labels": 2,
                "labels": {
                    "1": {
                        "count": 1024
                    },
                    "0": {
                        "count": 1024
                    }
                }
            }
        }
    },
    "train": {
        "num_samples": 4096,
        "number_of_characters": 421198,
        "number_texts_intersect_with_train": null,
        "min_text_length": 14,
        "average_text_length": 102.83154296875,
        "max_text_length": 402,
        "unique_text": 4096,
        "unique_labels": 2,
        "labels": {
            "1": {
                "count": 2048
            },
            "0": {
                "count": 2048
            }
        },
        "hf_subset_descriptive_stats": {
            "Danish": {
                "num_samples": 1024,
                "number_of_characters": 110271,
                "number_texts_intersect_with_train": null,
                "min_text_length": 14,
                "average_text_length": 107.6865234375,
                "max_text_length": 392,
                "unique_text": 1024,
                "unique_labels": 2,
                "labels": {
                    "1": {
                        "count": 512
                    },
                    "0": {
                        "count": 512
                    }
                }
            },
            "Norwegian_b": {
                "num_samples": 1024,
                "number_of_characters": 97878,
                "number_texts_intersect_with_train": null,
                "min_text_length": 18,
                "average_text_length": 95.583984375,
                "max_text_length": 350,
                "unique_text": 1024,
                "unique_labels": 2,
                "labels": {
                    "1": {
                        "count": 512
                    },
                    "0": {
                        "count": 512
                    }
                }
            },
            "Norwegian_n": {
                "num_samples": 1024,
                "number_of_characters": 107913,
                "number_texts_intersect_with_train": null,
                "min_text_length": 20,
                "average_text_length": 105.3837890625,
                "max_text_length": 402,
                "unique_text": 1024,
                "unique_labels": 2,
                "labels": {
                    "1": {
                        "count": 512
                    },
                    "0": {
                        "count": 512
                    }
                }
            },
            "Swedish": {
                "num_samples": 1024,
                "number_of_characters": 105136,
                "number_texts_intersect_with_train": null,
                "min_text_length": 19,
                "average_text_length": 102.671875,
                "max_text_length": 326,
                "unique_text": 1024,
                "unique_labels": 2,
                "labels": {
                    "1": {
                        "count": 512
                    },
                    "0": {
                        "count": 512
                    }
                }
            }
        }
    }
}

---

This dataset card was automatically generated using MTEB