Year: 2024
Type: conference-paper
Abstract: Social media anxiety detection refers to the approach of recognizing various signs of anxiety in a person based on their social media engagements. A person who experiences anxiety may express negative emotional responses like frustration, apprehension, and nervousness. Often, an individual's emotional tone in social media posts resonates with their feelings, providing valuable insights into their mental condition. The increase in socioeconomic adversities has resulted in a rapid growth of emotional disturbances in society. Identifying these emotional disturbances has become vital necessity for obtaining timely support and fostering better overall well-being by addressing the mental health component of the definition of health. Adapting pretrained language models (PLMs) for downstream tasks like digital mental health has therefore become a current trend even though there is no specific model for anxiety detection. There are many approaches for enhancing PLMs to determine downstream performance. In this research, we introduce a novel model, AnxiBERT, consists of soft prompt engineering logic, to adapt a PLMs knowledge for anxiety detection. To our knowledge, this is the first attempt to adapt a PLM's knowledge for anxiety detection using prompt engineering. Multiple evaluations show that our model demonstrates state-of-the-art performance compared to existing baselines.
Institution: University of Kelaniya
Language: English
Topic: Mental Health via Writing
Subfield: Social Psychology
Field: Psychology
Domain: Social Sciences
Open Access status: bronze
OpenAlex ID: W4409403870