ABSTRACT
The drive to automate information services has required the rapid advance of computer-aided language translation in many applications such as the translating phone. We present an Experimental Language Translation System for the Air Traffic Information Service(ATIS) in this paper. It can be used to translate English into Mandarin Chinese or vice versa in the ATIS domain. The headed phrase model and associated probabilistic finite state machines are used in our system. The dependency trees are introduced for representing phrases in both English and Chinese. We developed lexical transfer rules required for mapping analyzed English phrases into corresponding expressions in Mandarin Chinese. By evaluating the effectiveness of our approach against a random sample of test data, we found that the fluent and correct translation rate reaches 94%.
