Raj, Anuraag and Ali, Zain and Chaudhry, Shonal and Sharma, Anuraganand (2025) Uncovering Depression with LSTM and NLP Transformers in Social Media Posts. In: Data Science and Applications. Lecture Notes in Networks and Systems, 1265 . Springer Nature, Singapore, pp. 467-479. ISBN 978-981-96-2298-6
Full text not available from this repository. (Request a copy)Abstract
Depression detection from social media has garnered significant attention due to its potential to provide early intervention and support for individuals experiencing mental health challenges. In this study, we present a comprehensive comparative evaluation of the two most prominent state-of-the-art deep learning techniques for depression detection: Long Short-Term Memory (LSTM) networks and Natural Language Processing (NLP) Transformers. Leveraging data from social media platforms, we explore the efficacy of LSTM and NLP Transformer models in discerning patterns indicative of depressive symptoms. Our investigation involves preprocessing techniques tailored for social media text, feature extraction methodologies, and model architectures optimized for depression detection tasks. The comparative analysis shows that LSTM models achieve a precision of 0.97 and accuracy of 0.97%, which is better than Bidirectional Encoder Representations from Transformers (BERT). BERT achieves a precision of 0.95 and accuracy of 0.95% in detecting depression and non-depression states. These results suggest that LSTM models might be better suited for this task.
| Item Type: | Book Chapter |
|---|---|
| Additional Information: | Program Schedule at: https://scrs.in/conference/icdsa2024/page/ICDSA_2024%20Program%20Schedule |
| Uncontrolled Keywords: | Long Short-Term Memory |
| Subjects: | Q Science > QA Mathematics > QA75 Electronic computers. Computer science |
| Divisions: | School of Information Technology, Engineering, Mathematics and Physics (STEMP) |
| Depositing User: | Anuraganand Sharma |
| Date Deposited: | 22 Jul 2025 00:14 |
| Last Modified: | 07 Aug 2025 00:17 |
| URI: | https://repository.usp.ac.fj/id/eprint/14799 |
Actions (login required)
 |
View Item |