Enhancing Emotion Detection in Social Media with a Generative AI Approach to Analyzing Twitter Reviews
Vikram Nitin, N Enoch Das, E Meghana Sahithi, T. Santhosh, K. Tarun Sai · 2025
This study presents an automated approach to emotion detection and classification in informal text using SpaCy, with a focus on social media content such as tweets. The inherent brevity and colloquial nature of platforms like Twitter pose significant challenges for emotion analysis. To address this, the methodology incorporates web scraping of user reviews, each tagged with a unique identifier, and utilizes a Generative AI model to extract structured insights including dish names, ratings, pros, and cons in Markdown format. The analysis is conducted on the AIT-2018 dataset, enhanced with lexical resources such as WordNet-Affect and EmoSenticNet, enabling more nuanced emotion recognition. Experimental results demonstrate that the proposed model significantly outperforms traditional methods, particularly in handling short-form social media text. The findings underscore the effectiveness of automated systems in sentiment analysis and affective computing, offering valuable advancements in understanding user emotions and improving human-computer interaction.