Real-Time Mood Detection Using Tweets: A Hybrid Approach Combining BERT and Recurrent Neural Networks

Abstract

In the digital age, social media platforms like Twitter provide a rich source of textual data reflecting public sentiments and moods in real-time. This paper explores a hybrid approach to mood detection from tweets by combining Bidirectional Encoder Representations from Transformers (BERT) with Recurrent Neural Networks (RNNs). The proposed model leverages BERT's contextual embeddings and RNN's sequential processing capabilities to enhance mood detection accuracy. Experimental results on a benchmark Twitter sentiment dataset demonstrate that our hybrid model outperforms conventional models in terms of accuracy and F1 score, making it a viable solution for real-time mood analysis in social media.