# Emotion Detection and Analysis on Social Media

**Authors:** Bharat Gaind, Varun Syal, Sneha Padgalwar  
**arXiv:** [1901.08458](https://arxiv.org/abs/1901.08458)  
**Date:** 2019-01-24  
**Summary page:** https://www.summarizepaper.com/en/paper/1901.08458v2/

## Key points

- Authors address the challenge of detecting, classifying, and quantifying emotions in text from social media platforms
- Social media platforms are rich sources of emotions, feelings, and opinions expressed by people worldwide
- Proposed method to classify text into six different emotion categories: Happiness, Sadness, Fear, Anger, Surprise and Disgust
- Two approaches used: Natural Language Processing techniques and Machine Learning classification algorithms
- Automated method developed for creating the training set itself, eliminating the need for manual annotation of large datasets
- Comprehensive collection of emotional words along with their corresponding emotion intensities created
- Model accuracy in classifying tweets collected from Twitter tested and shown to be significantly high
- Research contributes to advancing sentiment analysis techniques by addressing the specific challenge of emotion classification in text data from social media sources.

## Summary

In their paper titled "Emotion Detection and Analysis on Social Media," authors Bharat Gaind, Varun Syal, and Sneha Padgalwar address the challenge of detecting, classifying, and quantifying emotions in text from social media platforms like Twitter. The authors emphasize that social media platforms are rich sources of emotions, feelings, and opinions expressed by people worldwide. To tackle this problem, the authors propose a method to classify text into six different emotion categories: Happiness, Sadness, Fear, Anger, Surprise and Disgust. They employ two approaches in their model to effectively extract these emotions from text. The first approach utilizes Natural Language Processing techniques and incorporates various textual features such as emoticons, degree words and negations; Parts Of Speech analysis; and other grammatical analyses. The second approach relies on Machine Learning classification algorithms. One notable contribution of the paper is the development of an automated method for creating the training set itself which eliminates the need for manual annotation of large datasets. Additionally, the authors have created a comprehensive collection of emotional words along with their corresponding emotion intensities. The accuracy of their model in classifying tweets collected from Twitter has been tested and shown to be significantly high. The findings presented in this paper provide valuable insights into emotion detection and analysis on social media platforms like Twitter. Overall, this research contributes to advancing sentiment analysis techniques by addressing the specific challenge of emotion classification in text data from social media sources.