Sanjana Kavatagi


2023

pdf bib
IIITDWD@LT-EDI-2023 Unveiling Depression: Using pre-trained language models for Harnessing Domain-Specific Features and Context Information
Shankar Biradar | Sunil Saumya | Sanjana Kavatagi
Proceedings of the Third Workshop on Language Technology for Equality, Diversity and Inclusion

Depression has become a common health problem impacting millions of individuals globally. Workplace stress and an unhealthy lifestyle have increased in recent years, leading to an increase in the number of people experiencing depressive symptoms. The spread of the epidemic has further exacerbated the problem. Early detection and precise prediction of depression are critical for early intervention and support for individuals at risk. However, due to the social stigma associated with the illness, many people are afraid to consult healthcare specialists, making early detection practically impossible. As a result, alternative strategies for depression prediction are being investigated, one of which is analyzing users’ social media posting behaviour. The organizers of LT-EDI@RANLP carried out a shared Task to encourage research in this area. Our team participated in the shared task and secured 21st rank with a macro F1 score 0f 0.36. This article provides a summary of the model presented in the shared task.