The Machine Translation of Noisy Text (MTNT) dataset is a Machine Translation dataset that consists of noisy comments on Reddit and professionally sourced translation. The translation are between French, Japanese and French, with between 7k and 37k sentence per language pair.
51 PAPERS • NO BENCHMARKS YET
The Multilingual Reuters Collection dataset comprises over 11,000 articles from six classes in five languages, i.e., English (E), French (F), German (G), Italian (I), and Spanish (S).
13 PAPERS • 1 BENCHMARK
The Medical Translation Task of WMT 2014 addresses the problem of domain-specific and genre-specific machine translation. The task is split into two subtasks: summary translation, focused on translation of sentences from summaries of medical articles, and query translation, focused on translation of queries entered by users into medical information search engines. Both subtasks included translation between English and Czech, German, and French, in both directions.
1 PAPER • NO BENCHMARKS YET