Recovering from Chinese-English machine translation parser failures

The goal of the analysis stage of a Chinese-English machine translation is produce a correct tree. A correct tree may not be guaranteed to produce for any sentence when there is the wrong input, or excluded vocabulary words, or a wrong rule, etc. It is very important for machine translation to be able to recover from partial analytical results. The recovery process from partial analytical results is presented using a hybrid statistical-linguistic based approach.