metadata

dataset_info:
  features:
    - name: lp
      dtype: large_string
    - name: src
      dtype: large_string
    - name: mt
      dtype: large_string
    - name: ref
      dtype: large_string
    - name: raw
      dtype: float64
    - name: domain
      dtype: large_string
    - name: year
      dtype: int64
    - name: sents
      dtype: int32
  splits:
    - name: train
      num_bytes: 36666470784
      num_examples: 7650287
    - name: test
      num_bytes: 283829719
      num_examples: 59235
  download_size: 23178699933
  dataset_size: 36950300503
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: test
        path: data/test-*
license: apache-2.0
size_categories:
  - 1M<n<10M
language:
  - bn
  - cs
  - de
  - en
  - et
  - fi
  - fr
  - gu
  - ha
  - hi
  - is
  - ja
  - kk
  - km
  - lt
  - lv
  - pl
  - ps
  - ru
  - ta
  - tr
  - uk
  - xh
  - zh
  - zu
tags:
  - mt-evaluation
  - WMT
  - 41-lang-pairs

Dataset Summary

Long-context / document-level dataset for Quality Estimation of Machine Translation. It is an augmented variant of the sentence-level WMT DA Human Evaluation dataset. In addition to individual sentences, it contains augmentations of 2, 4, 8, 16, and 32 sentences, among each language pair lp and domain. The raw column represents a weighted average of scores of augmented sentences using character lengths of src and mt as weights. The code used to apply the augmentation can be found here.

This dataset contains all DA human annotations from previous WMT News Translation shared tasks. It extends the sentence-level dataset RicardoRei/wmt-da-human-evaluation, split into train and test. Moreover, the raw column is normalized to be between 0 and 1 using this function.

The data is organised into 8 columns:

lp: language pair
src: input text
mt: translation
ref: reference translation
raw: direct assessment
domain: domain of the input text (e.g. news)
year: collection year
sents: number of sentences in the text

You can also find the original data for each year in the results section: https://www.statmt.org/wmt{YEAR}/results.html e.g: for 2020 data: https://www.statmt.org/wmt20/results.html

Python usage:

from datasets import load_dataset
dataset = load_dataset("ymoslem/wmt-da-human-evaluation-long-context")

There is no standard train/test split for this dataset, but you can easily split it according to year, language pair or domain. e.g.:

# split by year
data = dataset.filter(lambda example: example["year"] == 2022)

# split by LP
data = dataset.filter(lambda example: example["lp"] == "en-de")

# split by domain
data = dataset.filter(lambda example: example["domain"] == "news")

Note that most data is from the News domain.

Citation Information

If you use this data please cite the WMT findings from previous years: