[libvoikko] Changes to VFST simple speller configuration

Harri Pitkänen hatapitk at iki.fi
Sun Nov 1 15:03:34 EET 2015


VFST format has now replaced Malaga as the enabled-by-default dictionary 
format for Finnish language on master branch of libvoikko. There have been 
other configuration changes as well such as enabling HFST format by default. I 
will write more about these in the near future.

Meanwhile I would like to mention one change. The simple (language 
independent) VFST dictionary format based on weighted VFST transducers used to 
live as an experimental extension to dictionary format 2. Now that Malaga is 
disabled by default it is better to place it under dictionary format 5 
instead. The change is simple and the resulting configuration is cleaner than 
it used to be:

- Move transducers under some_path/2/mor-lang/*.vfst to some_path/5/mor-
lang/*.vfst (only the format version number changes)

- voikko-fi_FI.pro is no longer needed. Instead of that you need the following 
file content under some_path/5/mor-lang/index.txt:

Voikko-Dictionary-Format: 5
Language: fi-x-svfst
Description: Integration test dictionary for simple VFST dictionary backend
Morphology-Backend: vfst
Speller-Backend: vfst
Suggestion-Backend: vfst
Grammar-Backend: null

(Language and Description should of course be changed to match the actual 


More information about the Libvoikko mailing list