@GMarzinotto it's a good suggestion. But the bulk of the pywsd still relies on the various lesk algorithm. To extend the code to other languages we have to first get translations for the glosses (i.e. definitions) for every Synset.

@alvations
Question: how to generate the Synset and glosses?

@geekan Do you mean "how to generate synset and glosses" for other languages? If so, then the first thing is to take synsets from Open Multilingual Wordnet (OMW) that maps the synsets from the other languages to the princeton IDs then somehow translate the glosses from English to the other languages (manually/automatically).

Just a straw poll, is scaling to other languages more important to implementing more state-of-art algorithms for English? If so, BabelNet would be an option, but it's non-commerical =(