Emotion Detection From Social Media Posts

TL;DR

This paper tackles emotion detection in Twitter posts by comparing traditional machine learning methods with deep neural networks for classifying text into Fear, Anger, Joy, and Sadness. It systematically evaluates feature extraction techniques (Bag-of-Wwords vs. pre-trained word embeddings) and multiple architectures, finding BiGRU and an ensemble of BiGRU+BiLSTM to deliver the best performance at $87.66\%$ accuracy. Key contributions include a detailed hyperparameter tuning approach, architecture-specific design choices (CNN for sentence classification and recurrent networks for sequential modeling), and an analysis of class-wise performance and ROC metrics. The results demonstrate the effectiveness of embedding-based deep models for emotion detection in social media, with practical implications for tools that visualize emotional fluctuations and support targeted content recommendations, while pointing to future work in larger datasets, transfer learning, and graph-based models.

Abstract

Over the last few years, social media has evolved into a medium for expressing personal views, emotions, and even business and political proposals, recommendations, and advertisements. We address the topic of identifying emotions from text data obtained from social media posts like Twitter in this research. We have deployed different traditional machine learning techniques such as Support Vector Machines (SVM), Naive Bayes, Decision Trees, and Random Forest, as well as deep neural network models such as LSTM, CNN, GRU, BiLSTM, BiGRU to classify these tweets into four emotion categories (Fear, Anger, Joy, and Sadness). Furthermore, we have constructed a BiLSTM and BiGRU ensemble model. The evaluation result shows that the deep neural network models(BiGRU, to be specific) produce the most promising results compared to traditional machine learning models, with an 87.53 % accuracy rate. The ensemble model performs even better (87.66 %), albeit the difference is not significant. This result will aid in the development of a decision-making tool that visualizes emotional fluctuations.

Figures (5)

• Figure 1: Architecture for Convolutional Neural Network (CNN)
• Figure 2: Architecture for LSTM/GRU/BiLSTM/BiGRU
• Figure 3: Accuracy and loss curve of the Ensemble model
• Figure 4: Confusion matrix of the ensemble model
• Figure 5: ROC curve of the ensemble model