时间：2017年3月29日（星期三）上午9:00

教室：2#-209
 

题目：Neural Network based method for Affective ratings Prediction
摘要：A large number of research have thrown into the depression mining, mainstream of them is a category based approach, that is simply represents user opinion as several discrete classes such as positive or negative. few of them have focused on the dimensional technology, which represents the opinion as continuous numerical values on multiple dimensions, such as the Valence-Arousal (VA) space, which provide more fine-grained sentiment analysis. this paper first manually annotated a set of mental health corpus with Valence-Arousal rating, and then compare the predict effection of lexicon-based method to word embedding with deep learning approach base on this corpus, finally, an actual case demonstrated the change of mood of the users in social network，this change was tracked through the dimension technology, shown the significance of this method in practical application.