README.md

metadata

language:
  - de
  - en
  - es
  - fr
  - it
  - ja
  - nl
  - pt
  - ru
  - sv
  - multilingual
configs:
  - config_name: French
    data_files:
      - split: train
        path: train/fr/train.tsv
      - split: validation
        path: valid/fr/valid.tsv
      - split: test
        path: test/fr/test.tsv
  - config_name: German
    data_files:
      - split: train
        path: train/de/train.tsv
      - split: validation
        path: valid/de/valid.tsv
      - split: test
        path: test/de/test.tsv
  - config_name: English
    data_files:
      - split: train
        path: train/en/train.tsv
      - split: validation
        path: valid/en/valid.tsv
      - split: test
        path: test/en/test.tsv
  - config_name: Spanish
    data_files:
      - split: train
        path: train/es/train.tsv
      - split: validation
        path: valid/es/valid.tsv
      - split: test
        path: test/es/test.tsv
  - config_name: Italian
    data_files:
      - split: train
        path: train/it/train.tsv
      - split: validation
        path: valid/it/valid.tsv
      - split: test
        path: test/it/test.tsv
  - config_name: Japanese
    data_files:
      - split: train
        path: train/ja/train.tsv
      - split: validation
        path: valid/ja/valid.tsv
      - split: test
        path: test/ja/test.tsv
  - config_name: Dutch
    data_files:
      - split: train
        path: train/nl/train.tsv
      - split: validation
        path: valid/nl/valid.tsv
      - split: test
        path: test/nl/test.tsv
  - config_name: Portuguese
    data_files:
      - split: train
        path: train/pt/train.tsv
      - split: validation
        path: valid/pt/valid.tsv
      - split: test
        path: test/pt/test.tsv
  - config_name: Russian
    data_files:
      - split: train
        path: train/ru/train.tsv
      - split: validation
        path: valid/ru/valid.tsv
      - split: test
        path: test/ru/test.tsv
  - config_name: Swedish
    data_files:
      - split: train
        path: train/sv/train.tsv
      - split: validation
        path: valid/sv/valid.tsv
      - split: test
        path: test/sv/test.tsv

Dataset origin: https://github.com/zhangmozhi/mrs/

Multilingual Reply Suggestion (MRS)

MRS is a multilingual dataset for automatic reply suggestion. It is also an interesting benchmark for cross-lingual generalization.

Citation

If you find MRS useful, please cite:

@inproceedings{zhang-2021-mrs,
    title = {A Dataset and Baselines for Multilingual Reply Suggestion},
    author = {Mozhi Zhang and Wei Wang and Budhaditya Deb and Guoqing Zheng and Milad Shokouhi and Ahmed Hassan Awadallah},
    booktitle = {Proceedings of the Association for Computational Linguistics},
    doi = "10.18653/v1/2021.acl-long.97",
    year = {2021}
}