An effective way of word-level language identification for code-mixed facebook comments using word-embedding via character-embedding.