【摘要】：正Research in opinion analysis have drawn a great attention these days.Many of the effective opinion analysis system are based on supervised learning technology.However there lack of annotation sentiment corpora for Chinese opinion analysis.The purpose of our work is try to make use of annotation English corpora,where are rich and reliable to improve opinion analysis in Chinese.We propose a approach that to construct Chinese corpora by translating English corpora using different translation engines,as well as introducing domain adaption technique and meta-learning framework in order to improve the accuracy and stability of identifying subjective sentences.Through comparative evaluation with upper bound, we show the effectiveness of the proposed method.