How to piss off an orthographist(twitlyt.com) |
How to piss off an orthographist(twitlyt.com) |
I wrote up Twitlyt (http://www.twitlyt.com).
The algorithm is based on the assumption that, as long as the crammed word sounds like the actual word, it is comprehensible. I have used Python's Fuzzy library to generate soundex codes of each of the words in a sentence, then, I removed vowels from the actual words till I get the shortest substring(?) of the actual word that shares the same soundex code with the actual word.
Right now, it handles URLs and mentions. Let me know what you guys think about it.