Abstracts Track 2023

Area 1 - Web Intelligence and Semantic Web

Nr: 81

Detecting the Severity of Human Depression in Multi-Party Text Conversations


Prasan Yapa, Zilu Liang and Ian Piumarta

Abstract: Early identification of the severity of human depression has become a vital necessity for the well-being of society. People tend to release their tension through posting, commenting and group chatting on social media platforms resulting in plenty of social media data for possible research opportunities such as quantifying mental health on social media. Social media multi-party conversation (MPC) analysis focusses on discovering complex discourse level structures among multiple utterances which allows modelling of specific downstream tasks like depression severity detection. Even though MPC analysis is a growing and challenging research area, satisfactory work remains to be done for quantifying mental health based on complex discourse-level structures among multiple utterances of an MPC. In this study, we propose our idea for detecting the severity of human depression in multi-party text conversations. To our knowledge, this is the first attempt at human depression severity detection by analyzing text-based social media MPCs. We strongly believe that this research work could establish a sustainable computational mechanism for early identification of the severity of human depression.