Leveraging Emotion-Specific features to improve Transformer performance for Emotion Classification

Shaily Desai, Atharva Kshirsagar, Aditi Sidnerlikar, Nikhil Khodake, Manisha Marathe


Abstract
This paper describes team PVG’s AI Club’s approach to the Emotion Classification shared task held at WASSA 2022. This Track 2 sub-task focuses on building models which can predict a multi-class emotion label based on essays from news articles where a person, group or another entity is affected. Baseline transformer models have been demonstrating good results on sequence classification tasks, and we aim to improve this performance with the help of ensembling techniques, and by leveraging two variations of emotion-specific representations. We observe better results than our baseline models and achieve an accuracy of 0.619 and a macro F1 score of 0.520 on the emotion classification task.
Anthology ID:
2022.wassa-1.24
Volume:
Proceedings of the 12th Workshop on Computational Approaches to Subjectivity, Sentiment & Social Media Analysis
Month:
May
Year:
2022
Address:
Dublin, Ireland
Venue:
WASSA
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
245–249
Language:
URL:
https://aclanthology.org/2022.wassa-1.24
DOI:
10.18653/v1/2022.wassa-1.24
Bibkey:
Cite (ACL):
Shaily Desai, Atharva Kshirsagar, Aditi Sidnerlikar, Nikhil Khodake, and Manisha Marathe. 2022. Leveraging Emotion-Specific features to improve Transformer performance for Emotion Classification. In Proceedings of the 12th Workshop on Computational Approaches to Subjectivity, Sentiment & Social Media Analysis, pages 245–249, Dublin, Ireland. Association for Computational Linguistics.
Cite (Informal):
Leveraging Emotion-Specific features to improve Transformer performance for Emotion Classification (Desai et al., WASSA 2022)
Copy Citation:
PDF:
https://aclanthology.org/2022.wassa-1.24.pdf
Video:
 https://aclanthology.org/2022.wassa-1.24.mp4