I'm looking to publish a number of Russian NLP tools in a single python module, using HFST/VISLCG3. The hfst python module works well, but I can't find a straightforward way to use vislcg3 from python.
I can, of course, convert the hfst output (tuples) into one of the accepted stream formats and send it to a subprocess (as is done here
), but accessing cg3 as a library would be more efficient.
I found this github repo
that seems promising, but it's not well documented, and I haven't had time to dig in to it. Does anyone know of example code using this module?