Hassles and Uplifts Detection on Social Media Narratives

ACL ARR 2025 February Submission2731 Authors

Abstract: Hassles and uplifts provide key psychological information about individuals' reactions to daily stressful situations. Identifying and collecting this information poses challenges that conventional sentiment analysis cannot fully resolve. To address this, we introduce a novel task called Hassles and Uplifts Detection (HUD) and benchmark various language models on a dataset sourced from a private social media platform. Our findings indicate that existing LLMs may not yet be fully reliable for HUD, as several key aspects require further attention. Additionally, we propose an approach to demonstrate the transferability of experimental results, overcoming the common challenge of directly publishing private datasets in the mental health domain.

Paper Type: Long

Research Area: NLP Applications

Research Area Keywords: mental health, social media analysis, classification

Contribution Types: NLP engineering experiment, Approaches to low-resource settings, Data analysis

Languages Studied: English

Submission Number: 2731