I was lounging about today, idly thinking about folksonomies (hey..it could happen) and I had what I consider a somewhat interesting idea. Are there any existing sites that allow for tagging in multiple languages? I suppose that del.icio.us does by default for language that use the Roman alphabet, but what about systems that use a non-roman…does flickr or technorati allow for Chinese or Japanese kanji? Or for Farsi?
For any system where this were the case, and there was an enormous database of folksonomic data to mine, and the folksonomies were in some way descriptive (it’s possible to have non-descriptive folksonomies…some people actually leverage del.icio.us by using specifically non-descriptive tags in order to pull very specific things from the organization)…well, if you were describing things in the world…would you be able to data mine such a folksonomy as a translation engine?
You would imagine that on flickr, a picture of a red ball might be tagged “red” and “Ball” by multiple languages. By doing some basic statistical work on the data, I think you could come up with a pretty good translation engine.
Anyone out there see that this couldn’t/wouldn’t work? Would this be better than traditional translation engines…I don’t know. It leverages the wisdom of crowds in an interesting new way, though.
