Sentiment Analysis in Social Media Using Deep Neural Models: A Comprehensive Review of Emotional Dynamics and Emerging Trends

This study gives a clear summary of ways to spot feelings and views on social media, explaining their move from simple keyword lists to advanced AI systems such as deep neural nets and multi-source data blends. Results come from 22 trustworthy articles selected carefully through rigorous screening in six major online databases. It shows modern machine learning tools - like combined CNN-RNN models, attention-based networks, transformers, and hybrid multimodal frameworks - boost precision when understanding meaning and sentiment tone. Most studies use Twitter information since its freely accessible and refreshed in real time. Still, challenges persist including sarcasm detection, multilingual differences, topic shifts, small labeled samples, along with ethical concerns about personal data and skewed predictions. The review highlights fresh options with knowledge-driven models and graph structures, incorporating text, audio, or visuals. Yet, the study emphasizes stable datasets, clear processing rules, along with cross-platform efficiency to boost dependable emotion recognition in social media contexts.