An Attention-based Approach to Detect Emotion from Tweets

Abstract

In today’s world, social networks are the place where user share their views, emotions in their way. Social media, such as Twitter, Instagram, Facebook, etc. where millions of people express their views in their daily day-to-day life, which can be their sentiments and opinions or expressing emotions about a particular thing or their own. This gave researchers an outstanding opportunity to analyze the emotions of users’ activities on social networks. These massive digital data contain people’s day to day life sentiments, opinions, and showing emotions. Over the years there have been different research on emotion analysis of the social platform. As people tend to have different thoughts, analyzing the right emotion from social data is becoming a challenge. This clearly states that there is a need for an attempt to work towards these problems and it has opened up several opportunities for future research for hidden emotion identification, users’ emotions about a particular topic, etc. Detecting emotion from text is one of the toughest challenges in natural language processing. Developing a system that can detect emotion from social media is a crying need as people are sharing more of their thoughts here. In this research work, to learn the representation of the tweets, we propose an attention-based model. The proposed model has been divided into different components and subcomponents consisting of ID Convolution, Bidirectional LSTM, and Attention mechanism. We create a new dataset from SemEval Affect in Tweets dataset and then conduct experiments for the best outcomes. Our model achieves up to 79% accuracy in this task.