About this project

This repository contains finite state transducer (FST) source files and language resources for the Komi-Zyrian language, developed as part of the GiellaLT project. The resources include morphological analysers, proofing tools, and dictionaries built using finite state technology and Constraint Grammar. The repository provides speller files in `.bhfst` format for both desktop and mobile platforms, which can be tested using the divvunspell tool. These speller files are the same ones installed on users' computers and mobile phones, with different error models for desktop and mobile use. Nightly and CI/CD installation packages are available for Windows, macOS, and mobile platforms, though these are not tested for language quality and may contain regressions. Key features include: - Finite state source files for Komi-Zyrian morphology - Morphological analysers and proofing tools - Dictionaries and language data - Speller files for desktop and mobile platforms - Integration with the Divvun Installer for easy installation of proofing tools and keyboards The project depends on FST compilers such as HFST, Foma, or Xerox Xfst, and VislCG3 Constraint Grammar tools. Installation instructions are provided for macOS, Ubuntu/Debian, and RedHat/Fedora systems, with one-liner commands available for each platform. The Apertium wiki also provides detailed installation guides. The repository follows the GNU build system with autogen.sh, configure, and make commands. The autogen.sh script automatically clones or checks out other GiellaLT dependencies. Documentation is available at the GiellaLT documentation site, and the project includes research questions and maturity classification badges showing lemma count and speller suggestion quality. The work has been published in academic literature, with a 2021 paper in the Proceedings of the Seventh International Workshop on Computational Linguistics of Uralic Languages describing the open-source morphology development for the Komi-Zyrian language. The paper is authored by Jack Rueter, Niko Partanen, Mika Hämäläinen, and Trond Trosterud. The data and implementation are licensed under the LICENSE specified in the repository, with authors available to grant other licensing choices. The repository also includes a companion keyboard repository for the Komi-Zyrian language.