Since 2010, WASSA has brought together researchers working on Subjectivity,
Sentiment Analysis, Emotion Detection and Classification and their applications
to other NLP or real-world tasks (e.g. public health messaging, fake news,
media impact analysis, social media mining, computational literary stud- ies)
and researchers working on interdisciplinary aspects of affect computation from
text. In the past years we have noticed that WASSA offers a platform to
researchers investigating senti- ment and emotion in lesser-resourced
languages. We continued these efforts in the 2024 edition by featuring a
Special Track on multilinguality and the social bridge between high- and
lesser-resourced languages/communities. However, this year the majority of
accepted papers (18 of 29) only perform experiments in English, another six
include English and other languages, and only seven are entirely non-English.
This emphasizes the need to continue promoting such a multilingual track. The
topics of this edition of WASSA range from complex sentiment relations
(entity-level, long doc- uments, long-term narratives), mitigating unwanted
subjectivity and biases in models, and advancing towards high-level social
implications of subjective NLP. The ubiquity of Large Language Models is also a
common theme among the papers. This year's edition again featured two shared
tasks: Shared-Task 1: Empathy Detection and Emotion Classification, which is
already the fourth run of this shared task using a new, unpublished extension
of the dataset used last year. SharedTask 2: Explainability of Cross-lingual
Emotion Detection, on the other hand, is the first shared task on on
explainability of cross-lingual emotion detection. For the main workshop, we
received 57 direct submissions, of which 29 papers were accepted (accep- tance
rate of 53%). For Shared Task 1 we received 12 system description paper
submissions, while for Shared Task 2 we received 10 papers. Furthermore, one
ARR and one Findings paper will be presented in the poster session. In total,
55 papers will be presented at the workshop. Following the tradition of the
last two years, we again decided to award the best paper. The winner of the
Best Paper Award of this year's WASSA is Context is Important in Depressive
Language: A Study of the Interaction Between the Sentiments and Linguistic
Markers in Reddit Discussions by Neha Sharma and Kairit Sirts . Simultaneously,
we wish to promote the work in the Special Track, and therefore decided to
award a Special Track Best Paper award to Loneliness Episodes: A Japanese
Dataset for Loneliness Detection and Analysis by Naoya Fujikawa, Nguyen Quang
Toan, Kazuhiro Ito, Shoko Wakamiya, and Eiji Aramaki. On top of the main
workshop and shared task presentations, we are happy to have an invited speaker
who agreed to give a hybrid keynote at WASSA 2024: Professor Debora Nozza from
Bocconi University. We would like to thank the ACL Organizers and Workshop
chairs for their help and support during the preparation. We also thank the
OpenReview support team for their technical support. Finally, we espe- cially
thank the program committee for the time and effort they spent on reviewing,
especially to those who were willing to perform emergency reviews. The quality
of the program is a reflection of this hard work.