Guilt by Association: Emotion Intensities in Lexical Representations

Shahab Raji, Gerard de Melo


Abstract
What do linguistic models reveal about the emotions associated with words? In this study, we consider the task of estimating word-level emotion intensity scores for specific emotions, exploring unsupervised, supervised, and finally a self-supervised method of extracting emotional associations from pretrained vectors and models. Overall, we find that linguistic models carry substantial potential for inducing fine-grained emotion intensity scores, showing a far higher correlation with human ground truth ratings than state-of-the-art emotion lexicons based on labeled data.
Anthology ID:
2021.emnlp-main.781
Volume:
Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing
Month:
November
Year:
2021
Address:
Online and Punta Cana, Dominican Republic
Editors:
Marie-Francine Moens, Xuanjing Huang, Lucia Specia, Scott Wen-tau Yih
Venue:
EMNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
9911–9917
Language:
URL:
https://aclanthology.org/2021.emnlp-main.781
DOI:
10.18653/v1/2021.emnlp-main.781
Bibkey:
Cite (ACL):
Shahab Raji and Gerard de Melo. 2021. Guilt by Association: Emotion Intensities in Lexical Representations. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 9911–9917, Online and Punta Cana, Dominican Republic. Association for Computational Linguistics.
Cite (Informal):
Guilt by Association: Emotion Intensities in Lexical Representations (Raji & de Melo, EMNLP 2021)
Copy Citation:
PDF:
https://aclanthology.org/2021.emnlp-main.781.pdf
Video:
 https://aclanthology.org/2021.emnlp-main.781.mp4