针对微博的倾向性分析问题,提出了一种基于三元词组模式的情感分类方法。该方法通过构造情感词典及微博的三元词组模式,对未标注语料自动进行情感评分并标注情感极性,然后使用自动标注的语料训练得到情感分类器。在测试集上的实验结果表明,使用无人工参与标注的训练语料达到了79.26%的测试正确率。 According to the tendency analysis of Weibo, this paper proposes a sentiment classification method based on the three-phrase model. The method constructs the sentiment dictionary and microblogging ternary phrase mode, automatically sentiment scores and annotation emotional polarity on the unlabeled corpus, and then uses the automatic annotation corpus training to get the emotion classifier. The experimental results on the test set show that the training corpus with no human participation marks achieves 79.26% of the test accuracy.