Our first system (symbolic) for metrical scansion in Galician, presented in the following work:
Pablo Ruiz Fabo, Pauline Moreau, Anxo Alonso Pérez. (2026). Automatic Metrical Scansion of Galician Poetry: First Results. In Proceedings of the 17th International Conference on Computational Processing of Portuguese (PROPOR 2026) - Vol. 1, pp. 994–1004, Salvador, Brazil. Association for Computational Linguistics. https://aclanthology.org/2026.propor-1.101/
A web interface using the system is deployed at https://prf2.org/gama/.
Newer version: A later version of the system, installable with pip and with a command line and Python interface, is at compellit/gama. It reproduces the published results (see its evaluation).
The system is in src. An example client that runs both preprocessing and scansion on a text is at src/run_clients.sh.
From src, run_prepro.py runs preprocessing and run_scan_eval.py runs scansion. The -h option shows available options.
Note: This repository corresponds to the initial submission. Some improvements done before the camera-ready are not here; they are in compellit/gama, which reproduces the camera-ready results.
Required packages are in requirements.txt.
To run the system, some large files that wouldn't fit without LFS need to be downloaded. The following command downloads them to the required directory (script must be executable):
./download_large_files.shTo get an example output, you can run run_clients.sh (from src, script must be executable). This runs preprocessing and scansion with the same options used for the paper. There is an example poem in the script already, but you can pass a file with your own text using the -p flag. The command (using the poem already in the script) is:
./run_clients.sh -r 300 This will create a directory called ../texts/run_300/ with the poem at ../texts/run_300/example_001.txt. Preprocessing output will be in ../texts/run_300/out_300 and scansion output will be at ../texts/run_300/out_300/scansion_output_300.tsv. The value of -r is a run-ID.
To run with your own poem (here at path ../texts/example_input_poem.txt, the command would be (using 301 as a run ID):
./run_clients.sh -r 301 -p ../texts/example_input_poem.txt