metadata
task_categories:
- text-generation
language:
- ja
tags:
- text
- tv
- movies
pretty_name: OpenSubtitles dataset in Japanese
size_categories:
- 100M<n<1B
The dataset contains (almost) the entire OpenSubtittles database for Japanese:
- Over 7000 tv shows and/or movies.
- The subtittles are human generated.
- The dataset has been parsed, cleaned and converted to UTF-8.
File contents:
OpenSubtitles.parquet: The text and the time data.
OpenSubtitles_meta.parquet: The existing metadata for each title.
OpenSubtitles-OA.parquet: The dataset coded with two columns SOURCE(the name of the movie/tv show), and TEXT (the subtittles) following the Open Assistant rules.
Both tables can be joined by the ID column. (The value can be NULL in the meta table).