Study On Quality Evaluation Of English-Chinese Machine Translation

Machine translation evaluation has come up with the machine translaitonstudies. Machine translation is as difficult as machine translation itself. Since the 1980s, both machine translation and its evaluation have become hot topics in research. The evaluation of translation quality is a key topic in this field, while the automation of quality evalution is the emphasis.The N-gram language model based machine translation evaluation method is proposed in this thesis, which is an automatic evaluation model. Both prototype experiment and system performance evaluation are carried out based on the N-gram model. The popular methods of BLEU and NIST are studied in this thesis, which are compared with other mainstream evaluation methods, such as error analysis and weighted scoring, separate grammer point test method etc.Experiment shows consistency between the N-gram model-based method, the human evaluation result, and other mainstream evaluation methods. The automatic methods give quite precise evaluation results.