Fine-Grained Emotion Detection from Microblog Data Using Advanced NLP And Machine Learning Techniques
Year: 2025
Type: article
Abstract: : Social media platforms like Twitter, Instagram, and Facebook have exploded with user-generated content, offering a goldmine of data for understanding emotions online. But detecting nuanced feelings—like joy, anger, or surprise—in short, informal posts is far harder than basic sentiment analysis (which just labels things as "positive" or "negative"). This paper introduces a smarter way to detect emotions in microblogs by blending cutting-edge NLP and machine learning. Our key innovation? A hybrid model that combines the deep contextual understanding of transformer-based models (like BERT) with emotion- specific classifiers. Unlike older methods, our system doesn’t just skim the surface—it picks up subtle emotional cues, even in messy, slang-filled posts. We also tackle real-world challenges: emojis, sarcasm, and ever-changing internet slang. By fine-tuning our model on a diverse dataset (covering emotions from disgust to fear), we outperform traditional tools, especially for tricky cases like mixed emotions in a single tweet. The results? More accurate emotion tracking for applications like mental health monitoring, brand sentiment analysis, and real-time social media trends.
Authors: P.Bhanu Siva Rama Krishna, Dr.K.Sailaja
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4410254087