Hugging Face's logo

Datasets:
sbintuitions
/
JEMHopQA

Tasks:
Question Answering
Modalities:
Text
Formats:
parquet
Languages:
Japanese
Size:
1K - 10K
Libraries:
Datasets
pandas
License:
Dataset card Data Studio Files Community
1
JEMHopQA / README.md
teruo6939's picture
teruo6939
Update README.md
adab589 verified
preview code
|
raw
history blame
8.48 kB
metadata
language:
  - ja
license: cc-by-sa-4.0
task_categories:
  - question-answering
dataset_info:
  - config_name: v1
    features:
      - name: qid
        dtype: string
      - name: type
        dtype: string
      - name: question
        dtype: string
      - name: answer
        dtype: string
      - name: derivations
        list:
          - name: '0'
            dtype: string
          - name: '1'
            dtype: string
          - name: '2'
            sequence: string
      - name: page_ids
        sequence: string
      - name: time_dependent
        dtype: bool
    splits:
      - name: train
        num_bytes: 336414
        num_examples: 1059
      - name: validation
        num_bytes: 37295
        num_examples: 120
    download_size: 210192
    dataset_size: 373709
  - config_name: v1-extended-answers
    features:
      - name: qid
        dtype: string
      - name: type
        dtype: string
      - name: question
        dtype: string
      - name: answers
        sequence: string
      - name: derivations
        list:
          - name: '0'
            dtype: string
          - name: '1'
            dtype: string
          - name: '2'
            sequence: string
      - name: page_ids
        sequence: string
      - name: time_dependent
        dtype: bool
    splits:
      - name: validation
        num_bytes: 37999
        num_examples: 120
    download_size: 28606
    dataset_size: 37999
  - config_name: v1.1
    features:
      - name: qid
        dtype: string
      - name: type
        dtype: string
      - name: question
        dtype: string
      - name: answer
        dtype: string
      - name: derivations
        list:
          - name: '0'
            dtype: string
          - name: '1'
            dtype: string
          - name: '2'
            sequence: string
      - name: page_ids
        sequence: string
      - name: time_dependent
        dtype: bool
    splits:
      - name: train
        num_bytes: 336500
        num_examples: 1059
      - name: validation
        num_bytes: 37357
        num_examples: 120
    download_size: 210184
    dataset_size: 373857
  - config_name: v1.1-extended-answers
    features:
      - name: qid
        dtype: string
      - name: type
        dtype: string
      - name: question
        dtype: string
      - name: answers
        sequence: string
      - name: derivations
        list:
          - name: '0'
            dtype: string
          - name: '1'
            dtype: string
          - name: '2'
            sequence: string
      - name: page_ids
        sequence: string
      - name: time_dependent
        dtype: bool
    splits:
      - name: validation
        num_bytes: 38061
        num_examples: 120
    download_size: 28624
    dataset_size: 38061
  - config_name: v1.2
    features:
      - name: qid
        dtype: string
      - name: type
        dtype: string
      - name: question
        dtype: string
      - name: answer
        dtype: string
      - name: derivations
        list:
          - name: '0'
            dtype: string
          - name: '1'
            dtype: string
          - name: '2'
            sequence: string
      - name: page_ids
        sequence: string
      - name: time_dependent
        dtype: bool
    splits:
      - name: train
        num_bytes: 336860
        num_examples: 1059
      - name: validation
        num_bytes: 37389
        num_examples: 120
    download_size: 609248
    dataset_size: 374249
  - config_name: v1.2-extended-answers
    features:
      - name: qid
        dtype: string
      - name: type
        dtype: string
      - name: question
        dtype: string
      - name: answers
        sequence: string
      - name: derivations
        list:
          - name: '0'
            dtype: string
          - name: '1'
            dtype: string
          - name: '2'
            sequence: string
      - name: page_ids
        sequence: string
      - name: time_dependent
        dtype: bool
    splits:
      - name: validation
        num_bytes: 38093
        num_examples: 120
    download_size: 28663
    dataset_size: 38093
configs:
  - config_name: v1
    data_files:
      - split: train
        path: v1/train-*
      - split: validation
        path: v1/validation-*
  - config_name: v1-extended-answers
    data_files:
      - split: validation
        path: v1-extended-answers/validation-*
  - config_name: v1.1
    data_files:
      - split: train
        path: v1.1/train-*
      - split: validation
        path: v1.1/validation-*
  - config_name: v1.1-extended-answers
    data_files:
      - split: validation
        path: v1.1-extended-answers/validation-*
  - config_name: v1.2
    data_files:
      - split: train
        path: v1.2/train-*
      - split: validation
        path: v1.2/validation-*
  - config_name: v1.2-extended-answers
    data_files:
      - split: validation
        path: v1.2-extended-answers/validation-*

評価スコアの再現性確保と SB Intuitions 修正版の公開用クローン

ソース: aiishii/JEMHopQA on GitHub

JEMHopQA

JEMHopQA (Japanese Explainable Multi-hop Question Answering)は、回答導出ステップの情報付きの日本語の根拠情報付きマルチホップQAデータセットです。質問 (Question)を入力として、回答 (Answer)と導出 (Derivation)を生成するタスクです。導出は導出ステップの集合で、半構造化されたエンティティ間の関係表現です。問題は、Wikipediaの２つの記事の情報をリンクさせて答える構成問題 (compositional)と2つの記事の情報を比較して答える比較問題 (comparison)が含まれています。

Licensing Information

Creative Commons Attribution Share Alike 4.0 International

Citation Information 

@inproceedings{ishii-etal-2024-jemhopqa-dataset,
    title = "{JEMH}op{QA}: Dataset for {J}apanese Explainable Multi-Hop Question Answering",
    author = "Ishii, Ai  and
      Inoue, Naoya  and
      Suzuki, Hisami  and
      Sekine, Satoshi",
    editor = "Calzolari, Nicoletta  and
      Kan, Min-Yen  and
      Hoste, Veronique  and
      Lenci, Alessandro  and
      Sakti, Sakriani  and
      Xue, Nianwen",
    booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)",
    month = may,
    year = "2024",
    address = "Torino, Italia",
    publisher = "ELRA and ICCL",
    url = "https://aclanthology.org/2024.lrec-main.831",
    pages = "9515--9525",
}

Subsets

v1

v1: JEMHopQA/corpus on GitHub

  • qid (str): Unique identifier for each entry in the dataset.
  • type (str): The category of the question ("comparison" or "compositional").
  • question (str): The text of the question.（未 NFKC 正規化）
  • answer (str): The correct answer to the question.（未 NFKC 正規化）
  • derivations (dict[str: Any]): Knowledge triples for reasoning used to arrive at the answer.（未 NFKC 正規化）
  • page_ids (list[str]): Identifiers for related Wikipedia pages.
  • time_dependent (bool): Indicates whether the question/answer is time-sensitive.

オリジナルの JEMHopQA の derivations は、
list[ list[str, str, list[str]], list[str, str, list[str]]]
という構造になっている。しかし1つのリスト内に str型と list型が混在できない形式の都合、ここでは
list[{"0": str, "1": str, "2": list[str]}, {"0": str, "1": str, "2": list[str]}]
という形に修正している。具体的には以下の通り。

オリジナル

"derivations": [
["ダンガンロンパ 希望の学園と絶望の高校生", "発売日", ["2010年11月25日"]],
["ファイナルファンタジーXIII", "発売日", ["2009年12月17日"]]
],

↓ 修正版（このリポジトリ）

"derivations": [
{"0”": "ダンガンロンパ 希望の学園と絶望の高校生", "1": "発売日", "2": ["2010年11月25日"]},
{"0": "ファイナルファンタジーXIII", "1": "発売日", "2": ["2009年12月17日"]}
],

v1-extended-answers

  • v1 の answer に別解を加え、answers (list[str]) に拡張したもの
    • e.g., "カリフォルニア州クパチーノ" -> ["カリフォルニア州クパチーノ", "アメリカ合衆国カリフォルニア州クパチーノ", "アメリカ合衆国カリフォルニア州クパティーノ", "カリフォルニア州クパティーノ"]
  • split: validation のみ

v1.1

v1.1: JEMHopQA/corpus_ver1.1 on GitHub

v1.1-extended-answers

  • v1.1 の answer に別解を加え、answers (list[str]) に拡張したもの
  • split: validation のみ

v1.2

v1.2: JEMHopQA/corpus_ver1.2 on GitHub

v1.2-extended-answers

  • v1.2 の answer に別解を加え、answers (list[str]) に拡張したもの
  • split: validation のみ