Predicting Emotion Labels for Chinese Microblog Texts

摘要: We describe an experiment into detecting emotions in texts on the Chinese microblog service Sina Weibo (www.weibo.com) using distant supervision via various author-supplied emotion labels (emoticons and smilies). Existing word segmentation tools proved unreliable; better accuracy was achieved character-based features. Higher-order n-grams to be useful Accuracy varied according label emotion: while smilies are used more often, emoticons reliable. Happiness is most accurately predicted emotion, with accuracies around 90 % both gold-standard labels. This approach works well achieves high for happiness anger, it less effective sadness, surprise, disgust fear, which also difficult human annotators detect.