Deep Learning-Based Sentiment and Emotion Analysis of Social Media Data to Identify Factors Affecting Healthy Food Choices in Urban Communities

Abstract

The increasing influence of social media on public perception has made it a powerful driver of dietary behavior in urban communities. Nevertheless, the abundance of unverified health information often obscures individuals’ ability to make informed food choices. This study proposes a deep learning-based framework to analyze sentiment and emotion from social media discourse in order to uncover the key factors affecting healthy food decisions in urban settings. By applying Natural Language Processing (NLP) techniques and advanced deep learning models to a large corpus of user-generated content, the research identifies significant patterns linking emotional expression with food-related decision-making. The results indicate that positive emotions, such as pride and satisfaction, are strongly associated with healthy food promotion, while negative emotions, including frustration, are predominantly tied to affordability, accessibility, and convenience issues. Among these, price and food quality emerge as the most critical determinants shaping consumer preferences. These findings underscore the importance of integrating emotional and socio-economic considerations into public health strategies. Beyond offering empirical insights, this study demonstrates the scalability and effectiveness of deep learning in extracting nuanced perspectives from unstructured social media data, thereby contributing a robust methodological approach for real-time public health monitoring and intervention design.