17063 Emotional Expression through Hashtags: Sentiment and Psychological Insights from #Kaburajadulu Using Natural Language Processing (NLP) and Large Language Models (LLMs)

R3a-1

Abstract

The viral spread of the hashtag #kaburajadulu reflects growing public discontent in Indonesia, particularly among younger generations, in response to mounting social, economic, and political pressures. However, the psychological dimensions of such digital expressions remain underexplored. This study aims to investigate the emotional and psychological expressions embedded in the #kaburajadulu discourse on X (formerly Twitter), focusing on how users articulate stress, frustration, and coping in a public digital space. A dataset of approximately 6,000 to 12,000 tweets posted between January and February 2025 will be collected using the Twitter API. Natural Language Processing (NLP) and Large Language Models (LLMs), specifically the OpenAI GPT-4o API, will be used to classify the sentiment (positive, neutral, or negative) and identify underlying psychological constructs such as escapism, avoidance coping, frustration, and burnout. A Chi-square test will assess whether there is a significant association between sentiment polarity and the presence of specific psychological constructs. By mapping the psychological landscape of digital discourse, this study seeks to advance understanding of how social media functions as an outlet for collective emotional expression and informal coping. The anticipated results aim to provide valuable implications for mental health monitoring, policy communication strategies, and digital well-being frameworks. This study is among the first to apply AI-powered tools to analyze psychological constructs in Indonesian social media discourse.