metadata

configs:
  - config_name: default
    data_files:
      - split: test
        path: data/test-*
      - split: dev
        path: data/dev-*
dataset_info:
  features:
    - name: uid
      dtype: int64
    - name: question_text
      dtype: string
    - name: query
      dtype: string
    - name: answer_text
      dtype: string
    - name: question_uris
      sequence: string
    - name: question_props
      sequence: string
    - name: answers
      list:
        - name: datatype
          dtype: string
        - name: label
          dtype: string
        - name: type
          dtype: string
        - name: value
          dtype: string
        - name: wd_names
          struct:
            - name: en
              sequence: string
            - name: ru
              sequence: string
        - name: wp_names
          sequence: string
        - name: xml:lang
          dtype: string
    - name: paragraphs_uids
      struct:
        - name: all_related
          sequence: int64
        - name: with_answer
          sequence: int64
    - name: tags
      sequence: string
    - name: RuBQ_version
      dtype: string
    - name: question_eng
      dtype: string
  splits:
    - name: test
      num_bytes: 1992076
      num_examples: 2330
    - name: dev
      num_bytes: 488914
      num_examples: 580
  download_size: 0
  dataset_size: 2480990
license: cc-by-sa-4.0
task_categories:
  - question-answering
language:
  - ru
  - en
tags:
  - qa
  - machine reading
source_datasets:
  - original
pretty_name: RuBQ 2.0
size_categories:
  - 1K<n<10K
paperswithcode_id: rubq

RuBQ 2.0

Dataset Description

Repository: https://github.com/vladislavneon/RuBQ/tree/master/RuBQ_2.0
Paper: RuBQ: A Russian Dataset for Question Answering over Wikidata

For training data see d0rj/RuBQ_2.0-paragraphs.