Reordering the translations from Chinese into English based on peer patterns

Translation reordering in machine translation was studied for improving the quality of translation output. A method for reordering the translations from Chinese into English based on peer patterns was put forward below: the correspondent relationships between Chinese word orders and English word orders were extracted from phrase translation pairs, and then, the peer patterns, obtained based on the combination of linguistics features and error-driven learning, were used to change the structures of Chinese parsing trees, making the translations accord with English word orders. The new method was used to reorder Chinese sentences of 500 Chinese-English bilingual sentence pairs in an experiment, and the results showed the crossing rate of word links degraded 10.56%, and the translation quality was increased because the Chinese sentences were reordered.