Joint Model Of Micro-blog Emotion Cause Detection And Emotion Recognition Of Emoticons Based On Bi-LSTM-CRF

With the rise of social networks,more and more people begin to present their opinions on various topics by using social platforms such as Sina micro-blogs,therefore a lot of social texts with emotional characteristics appear,these texts make the social emotion analysis task has become popular in Natural Language Processing.Emotion cause detection as a deep text emotion understanding has become a hot issue in the field of text emotion analysis,current research usually regard emotion cause detection and emotion recognition as two independent tasks,which easily leads to propagation of errors.In this paper,considering that emotion cause detection and emotion recognition are interactcd,and that the emoticons in the micro-blog text can usually express the emotion of the text,we propose a joint model of emotion cause detection and emotion recognition of emoticons based on Bi-LSTM-CRF model.This model can mark the reasons of the emotion expressed in the micro-blog text and identify the emotional categories contained in the emoticons.The main contributions are shown as follows:(1)Building the micro-blog dataset of emotion cause and emoticons,by crawling Sina micro-blog texts by crawlers,and manually annotate emotional categories of emoticons and clauses of emotion causes.(2)Formalizing emotion cause detection and emotion recognition into a unified sequence labeling problem,it makes full use of the interaction between emotion causes and emotions and simultaneously processes the two tasks based on Bi-LSTM-CRF model.The experimcntal results on the micro-blog dataset built above show that joint model of micro-blog emotion cause detection and emotion recognition of emoticons based on Bi-LSTM-CRF has a better performance on both tasks than serial model of the two tasks.