MANTIS at SMM4H’2022: Pre-Trained Language Models Meet a Suite of Psycholinguistic Features for the Detection of Self-Reported Chronic Stress

Sourabh Zanwar, Daniel Wiechmann, Yu Qiao, Elma Kerz


Abstract
This paper describes our submission to Social Media Mining for Health (SMM4H) 2022 Shared Task 8, aimed at detecting self-reported chronic stress on Twitter. Our approach leverages a pre-trained transformer model (RoBERTa) in combination with a Bidirectional Long Short-Term Memory (BiLSTM) network trained on a diverse set of psycholinguistic features. We handle the class imbalance issue in the training dataset by augmenting it by another dataset used for stress classification in social media.
Anthology ID:
2022.smm4h-1.5
Volume:
Proceedings of The Seventh Workshop on Social Media Mining for Health Applications, Workshop & Shared Task
Month:
October
Year:
2022
Address:
Gyeongju, Republic of Korea
Editors:
Graciela Gonzalez-Hernandez, Davy Weissenbacher
Venue:
SMM4H
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
16–18
Language:
URL:
https://aclanthology.org/2022.smm4h-1.5
DOI:
Bibkey:
Cite (ACL):
Sourabh Zanwar, Daniel Wiechmann, Yu Qiao, and Elma Kerz. 2022. MANTIS at SMM4H’2022: Pre-Trained Language Models Meet a Suite of Psycholinguistic Features for the Detection of Self-Reported Chronic Stress. In Proceedings of The Seventh Workshop on Social Media Mining for Health Applications, Workshop & Shared Task, pages 16–18, Gyeongju, Republic of Korea. Association for Computational Linguistics.
Cite (Informal):
MANTIS at SMM4H’2022: Pre-Trained Language Models Meet a Suite of Psycholinguistic Features for the Detection of Self-Reported Chronic Stress (Zanwar et al., SMM4H 2022)
Copy Citation:
PDF:
https://preview.aclanthology.org/proper-vol2-ingestion/2022.smm4h-1.5.pdf