Skip to content

Latest commit

 

History

History
37 lines (23 loc) · 979 Bytes

README.md

File metadata and controls

37 lines (23 loc) · 979 Bytes
Raw

Statistical Dictionary

This software translates a text using an statistical dictionary.

Dependencies

To generate IBM Model 1 alignments, mgiza is needed.

Usage

Alignments

IBM1Alignments.sh source_file target_file dest_dir

Where:

  • source_file is the source from the parallel training data.
  • target_file is the target from the parallel training data.
  • dest_dir is the directory in which to save the alignments.

Note: the script expects a variable ($GIZA) pointing towards mgiza's bin directory.

Text translation

SD.py -t text_file -a alignments

Where:

  • text_file if the text to translate.
  • alignments is the file created at the alignments step (dest_dir/alignmetns).

Run through Docker

Alternatively, if you want to run this software using Docker, you can have a look at this repo.